PaperPlayer biorxiv bioinformatics: Recent Episodes
PaperPlayer
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.28.514184v1?rss=1
Authors: Moreau, E., Hardiman, O., Heverin, M., O'Sullivan, D.
Abstract: Motivation: Literature-Based Discovery (LBD) aims to help researchers to identify relations between concepts which are worthy of further investigation by text-mining the biomedical literature. While the LBD literature is rich and the field is considered mature, standard practice in the evaluation of LBD methods is methodologically poor and has not progressed on par with the domain. The lack of properly designed and decent-sized benchmark dataset hinders the progress of the field and its development into applications usable by biomedical experts. Results: This work presents a method for mining past discoveries from the biomedical literature. It leverages the impact made by a discovery, using descriptive statistics to detect surges in the prevalence of a relation across time. This method allows the collection of a large amount of time-stamped discoveries which can be used for LBD evaluation or other applications. The validity of the method is tested against a baseline representing the state of the art "time sliced" method. Availability: The source data used in this article are publicly available. The implementation and the resulting data are published under open-source license (code: https://github.com/erwanm/medline-discoveries; datasets: https://zenodo.org/record/5888572). An online exploration tool is also provided at https://brainmend.adaptcentre.ie/.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513573v1?rss=1
Authors: BICCN Data Ecosytem Collaboration,, Hawrylycz, M. J., Martone, M. E., Hof, P. R., Lein, E. S., Regev, A., Ascoli, G. A. A., Bjaalie, J. G., Dong, H.-W., Ghosh, S. S., Gillis, J., Hertzano, R., Haynor, D. R., Kim, Y., Liu, Y., Miller, J. A., Mitra, P. P., Mukamel, E., Osumi-Sutherland, D., Peng, H., Ray, P. L., Sanchez, R., Ropelewski, A., Scheuermann, R. H., Tan, S. Z. K., Tickle, T., Tilgner, H., Varghese, M., Wester, B., White, O., Aevermann, B., Allemang, D., Ament, S., Athey, T. L., Baker, P. M., Baker, C., Baker, K. S., Bandrowski, A., Bishwakarma, P., Carr, A., Chen, M., Choudhury, R.,
Abstract: Characterizing cellular diversity at different levels of biological organization across data modalities is a prerequisite to understanding the function of cell types in the brain. Classification of neurons is also required to manipulate cell types in controlled ways, and to understand their variation and vulnerability in brain disorders. The BRAIN Initiative Cell Census Network (BICCN) is an integrated network of data generating centers, data archives and data standards developers, with the goal of systematic multimodal brain cell type profiling and characterization. Emphasis of the BICCN is on the whole mouse brain and demonstration of prototypes for human and non-human primate (NHP) brains. Here, we provide a guide to the cellular and spatial approaches employed, and to accessing and using the BICCN data and its extensive resources, including the BRAIN Cell Data Center (BCDC) which serves to manage and integrate data across the ecosystem. We illustrate the power of the BICCN data ecosystem through vignettes highlighting several BICCN analysis and visualization tools. Finally, we present emerging standards that have been developed or adopted by the BICCN toward FAIR (Wilkinson et al. 2016a) neuroscience. The combined BICCN ecosystem provides a comprehensive resource for the exploration and analysis of cell types in the brain.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.513881v1?rss=1
Authors: Fan, J., Khan, J., Pibiri, G. E., Patro, R.
Abstract: The reference indexing problem for k-mers is to pre-process a collection of reference genomic sequences [R] so that the position of all occurrences of any queried k-mer can be rapidly identified. An efficient and scalable solution to this problem is fundamental for many tasks in bioinformatics. In this work, we introduce the spectrum preserving tiling (SPT), a general representation of [R] that specifies how a set of tiles repeatedly occur to spell out the constituent reference sequences in [R]. By encoding the order and positions where tiles occur, SPTs enable the implementation and analysis of a general class of modular indexes. An index over a SPT decomposes the reference indexing problem for k-mers into: (1) a k-mer-to-tile mapping; and (2) a tile-to-occurrence mapping. Recently introduced work to construct and compactly index k-mer-sets can be used to efficiently implement the k-mer-to-tile mapping. However, implementing the tile-to-occurrence mapping remains prohibitively costly in terms of space. As reference collections become large, the space requirements of the tile-to-occurrence mapping dominates that of the k-mer-to-tile mapping since the former depends on the amount of total sequence while the latter depends on the number of unique k-mers in [R]. To address this, we introduce a class of sampling schemes for SPTs that trade off speed to reduce the size of the tile-to-reference mapping. We implement a practical index with these sampling schemes in the tool: pufferfish2. When indexing 30,000 bacterial genomes, pufferfish2 reduces the size of the tile-to-occurrence mapping from 86.3G to 34.6G while incurring only a 3.6x slowdown when querying k-mers from a sequenced readset. Availability: pufferfish2 implemented in Rust and available at https://github.com/COMBINE-lab/pufferfish2 .
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.514030v1?rss=1
Authors: Bouilhol, E., Lefevre, E., Barry, T., Levet, F., Beghin, A., Viasnoff, V., Galindo, X., Galland, R., Sibarita, J.-B., Nikolski, M.
Abstract: Automatic segmentation of nuclei in low-light microscopy images remains a difficult task, especially for high-throughput experiments where need for automation is strong. Low saliency of nuclei with respect to the background, variability of their intensity together with low signal-to-noise ratio in these images constitute a major challenge for mainstream algorithms of nuclei segmentation. In this work we introduce SalienceNet, an unsupervised deep learning-based method that uses the style transfer properties of cycleGAN to transform low saliency images into high saliency images, thus enabling accurate segmentation by downstream analysis methods, and that without need for any parameter tuning. We have acquired a novel dataset of organoid images with soSPIM, a microscopy technique that enables the acquisition of images in low-light conditions. Our experiments show that SalienceNet increased the saliency of these images up to the desired level. Moreover, we evaluated the impact of SalienceNet on segmentation for both Otsu thresholding and StarDist and have shown that enhancing nuclei with SalienceNet improved segmentation results using Otsu thresholding by 30% and using StarDist by 26% in terms of IOU when compared to segmentation of non-enhanced images. Together these results show that SalienceNet can be used as a common preprocessing step to automate nuclei segmentation pipelines for low-light microscopy images.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.514069v1?rss=1
Authors: Egorov, A. A., Atkinson, G. C.
Abstract: Upstream open reading frames (uORFs, encoding so-called leader peptides) can regulate translation and transcription of downstream main ORFs (mORFs) in prokaryotes and eukaryotes. However, annotation of novel functional uORFs is challenging due their short size of usually less than 100 codons. While transcription- and translation-level next generation sequencing (NGS) methods can be used for genome-wide uORF identification, this data is not available for the vast majority of species with sequenced genomes. At the same time, the exponentially increasing amount of genome assemblies gives us the opportunity to take advantage of evolutionary conservation in our predictions of ORFs. Here we present a tool for conserved uORF annotation in 5' upstream sequences of a user-defined protein of interest or a set of protein homologues. It can also be used to find small ORFs within a set of nucleotide sequences. The output includes publication-quality figures with multiple sequence alignments, sequence logos and locus annotation of the predicted uORFs in graphical vector format. uORF4u is written in Python3 and runs on Linux and MacOS. The command-line interface covers most practical use cases, while the provided Python API allows usage within a Python program and additional customisation. Source code is available from the GitHub page: https://github.com/art-egorov/uorf4u. Detailed documentation that includes an example-driven guide available at the software home page: https://art-egorov.github.io/uorf4u.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.513962v1?rss=1
Authors: Avila Cartes, J., Bonizzoni, P., Ciccolella, S., Della Vedova, G., Denti, L., Monti, D., Pirola, Y., Porto, F.
Abstract: The transition towards graph pangenomes is posing several new challenging questions, most notably how to extend the classical notion of read alignment from a sequence-to-sequence to a sequence-to-graph setting. Especially on variation graphs, where paths corresponding to individual genomes are labeled, notions of alignments that are strongly inspired by the classical ones are usually able to capture only variations that can be expressed by mismatches or gaps, such as SNPs or short insertions and deletions. On the other hand the recent investigation of pangenomes at bacterial scale (Colquhoun et al, 2021) shows that most tools are tailored for human pangenomes and are not suited to bacteria which exhibit, among other characteristics, a larger variability. Such variability leads to the need for incorporating a greater flexibility when computing an alignment. In this paper, we extend the usual notion of sequence-to-graph alignment by including recombinations among the variations that explicitly represented and evaluated in an alignment. From a computational modeling point of view, a recombination corresponds to identifying a new path of the variation graph which is a mosaic of two different paths, possibly joined by a new arc. We provide a dynamic programming algorithm for computing an optimal alignment that allows recombinations with an affine penalty. We have implemented our approach with the tool RecGraph and we have analyzed its accuracy over some over some bacterial pangenome graphs.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513838v1?rss=1
Authors: Nagy, S. A., Kilim, O., Csabai, I., György, G., Solymosi, N.
Abstract: Body condition scoring is a simple method to estimate the energy supply of dairy cattle. Our study aimed to investigate the accuracy with which supervised machine learning, a deep convolutional neural network, can be used to retrieve body condition score (BCS) classes estimated by an expert. Using a simple action camera, we recorded images of animals' rumps in three large-scale farms. The images were annotated with three different-sized boxes by an expert. A Faster-RCNN pre-trained model was trained on 12 and 3 BCS classes. Training in 12 classes, with a 0 error range, the Cohen's kappa value yielded minimal agreement. Allowing an error range of 0.25, we obtained a minimum or week agreement. With an error range of 0.5, we had strong or almost perfect agreements. The kappa values of the approach trained on 3 classes show that we can classify all animals into BCS categories with at least moderate agreement. Furthermore, CNNs trained in 3 BCS classes show a remarkably higher proportion of strong agreement than those trained in 12 classes. The prediction precision based on training with various annotation regions showed no meaningful differences.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.514020v1?rss=1
Authors: Meysman, P., Barton, J., Bravi, B., Cohen-Lavi, L., Karnaukhov, V., Lilleskov, E., Montemurro, A., Nielsen, M., Mora, T., Pereira, P., Postovskaya, A., Rodriguez Martinez, M., Fernandez-de-Cossio-Diaz, J., Vujkovic, A., Walczak, A., Weber, A., Yin, R., Eugster, A., Sharma, V.
Abstract: Many different solutions to predicting the cognate epitope target of a T-cell receptor (TCR) have been proposed. However several questions on the advantages and disadvantages of these different approaches remain unresolved, as most methods have only been evaluated within the context of their initial publications and data sets. Here, we report the findings of the first public TCR-epitope prediction benchmark performed on 23 prediction models in the context of the ImmRep 2022 TCR-epitope specificity workshop. This benchmark revealed that the use of paired-chain alpha-beta, as well as CDR1/2 or V/J information, when available, improves classification obtained with CDR3 data, independent of the underlying approach. In addition, we found that straight-forward distance-based approaches can achieve a respectable performance when compared to more complex machine-learning models. Finally, we highlight the need for a truly independent follow-up benchmark and provide recommendations for the design of such a next benchmark.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.514084v1?rss=1
Authors: Hsu, J., Jarroux, J., Joglekar, A., Romero, J. P., Nemec, C., Reyes, D., Royall, A., He, Y., Belchikov, N., Leo, K., Taylor, S. E. B., Tilgner, H. U.
Abstract: Barcoding strategies are fundamental to droplet-based single-cell sequencing, and understanding the biases and caveats between approaches is essential. Here, we comprehensively evaluated both short and long reads of the cDNA obtained through the two marketed approaches from 10x Genomics, the "3 assay" and the "5 assay", which attach barcodes at different ends of the mRNA molecule. Although the barcode detection, cell-type identification, and gene expression profile are similar in both assays, the 5 assay captured more exonic molecules and fewer intronic molecules compared to the 3 assay. We found that 13.7% of genes sequenced have longer average read lengths and are more complete (spanning both polyA-site and TSS) in the long reads from the 5 assay compared to the 3 assay. These genes are characterized by long average transcript length, high intron number, and low expression overall. Despite these differences, cell-type-specific isoform profiles observed from the two assays remain highly correlated. This study provides a benchmark for choosing the single-cell assay for the intended research question, and insights regarding platform-specific biases to be mindful of when analyzing data, particularly across samples and technologies.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.28.514245v1?rss=1
Authors: Rakesh, M., Vezina, H., Laprise, C., Freeman, E., Burkett, K., Roy-Gagnon, M.-H.
Abstract: Summary Founder populations with deep genealogical data are well suited for investigating genetic variants contributing to diseases. Here, we present a new function added to the genealogical analysis R package GENLIB, which can simulate the transmission of haplotypes from founders to probands along very large and complex user-specified genealogies. Availability and implementation The new function is available in the latest version of the GENLIB package (v1.1.6), available on the CRAN repository and from https://github.com/R-GENLIB/GENLIB. Stand-alone scripts for analyzing the output of the function can be accessed at https://github.com/R-GENLIB/simuhaplo_scripts.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.28.514253v1?rss=1
Authors: Muehlberg, A., Ritter, P., Langer, S., Goossens, C., Nuebler, S., Schneidereit, D., Taubmann, O., Denzinger, F., Noerenberg, D., Haug, M., Goldmann, W., Maier, A., Friedrich, O., Kreiss, L.
Abstract: Deep learning (DL) shows notable success in biomedical studies. However, most DL algorithms work as a black box, exclude biomedical experts, and need extensive data. We introduce the Self-Enhancing Multi-Photon Artificial Intelligence (SEMPAI), that integrates hypothesis-driven priors in a data-driven DL approach for research on multiphoton microscopy (MPM) of muscle fibers. SEMPAI utilizes meta-learning to optimize prior integration, data representation, and neural network architecture simultaneously. This allows hypothesis testing and provides interpretable feedback about the origin of biological information in MPM images. SEMPAI performs joint learning of several tasks to enable prediction for small datasets. The method is applied on an extensive multi-study dataset resulting in the largest joint analysis of pathologies and function for single muscle fibers. SEMPAI outperforms state-of-the-art biomarkers in six of seven predictive tasks, including those with scarce data. SEMPAI's DL models with integrated priors are superior to those without priors and to prior-only machine learning approaches.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.28.514164v1?rss=1
Authors: Chuang, T.-J., Chiang, T.-W., Chen, C.-Y.
Abstract: Circular RNAs (circRNAs) are non-polyadenylated RNAs with a continuous loop structure characterized by a non-co-linear back-splice junction (BSJ). While dozens of computational tools have been developed and identified millions of circRNA candidates in diverse species, it remains a major challenge for determining circRNA reliability due to various types of false positives. Here, we systematically assess the impacts of numerous factors related to identification, conservation, biogenesis, and function on circRNA reliability by comparisons of circRNA expression from mock (total RNAs) and the corresponding co-linear/polyadenylated RNA-depleted datasets based on three different RNA treatment approaches. Eight important indicators of circRNA reliability are determined. The relative contribution to variability explained analyses further reveal that the relative importance of these factors in affecting circRNA reliability is conservation level of circRNA greater than full-length circular sequences greater than supporting BSJ read count greater than both BSJ donor and acceptor splice sites at the same co-linear transcript isoforms greater than both BSJ donor and acceptor splice sites at the annotated exon boundaries greater than BSJs detected by multiple tools greater than supporting functional features greater than both BSJ donor and acceptor splice sites undergoing alternative splicing. By extracting RT-independent circRNAs, circRNAs passing multiple experimental validations, and database-specific circRNAs, we showed the additive effects of these important factors in determining circRNA reliability. This study thus provides a useful guideline and an important resource for selecting high-confidence circRNAs for further investigations.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.513741v1?rss=1
Authors: Ant, L., Le Dily, F., Beato, M., Saragüeta, P.
Abstract: Decidualization is a differentiation process involving shape reorganization from a fibroblast to an epithelioid-like appearance of endometrial stromal cells. Specificities of these cells impede the use of existing automated tools to follow morphological changes during differentiation; we therefore developed a simple but accurate methodology to quantify the phenotypical changes that occur in an in vitro decidualization system. The approach consists of the analysis of the circularity of the cells directly from light microscopy images. Here, we used this methodology to follow the effects of progesterone or progestin R5020 in combination with estradiol (E2) and cAMP on inducing the decidualization of human endometrial cells. We further implemented a statistical model to detect the differences in the kinetics of decidualization of the two hormonal stimuli before all the cell population acquired the decidual phenotype. We found that 2 days after stimulation are sufficient to detect statistical differences in morphology between decidualization induced and control cells. Here, we detail the model and scripts in order to provide a useful, practical and low cost tool to evaluate morphological aspects of endometrial stromal differentiation.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.514080v1?rss=1
Authors: Yang, G., Huang, S., Hu, K., Lu, A., Yang, J., Meroueh, N., Dang, P., Zhu, H., Cao, S., Zhang, C.
Abstract: Glucose and glutamine are major carbon and energy sources that promote the fast proliferation of cancer cells. Metabolic shifs observed on cell line or mouse models may not reflect the general metabolic shifts in real human cancer tissue. In this study, we conducted a computational characterization of the flux distribution and variations of the central energy metabolism and key branches in a pan-cancer analysis, including glycolytic pathway, production of lactate, TCA cycle, nucleic acids synthesis, glutaminolysis, glutaminate, glutamine and glutathione metabolism, amino acids synthesis, in 11 cancer subtypes and 9 matched adjacent normal tissue types, by using TCGA tissue transcriptomics data. Our analysis confirms the increased influx in glucose uptake and glycolysis and decreased upper part of TCA cycle, i.e., Warbug effect in the analyzed cancer types. However, consistently increased lactate production and second half of TCA cycle were only seen in certain cancer types. More interestingly, we did not see cancer tissues have highly shifted glutaminolysis compared to their adjacent normal controls. A systems biology model of metabolic shifts in through cancer and tissue types is further developed and analyzed. We observed that (1) normal tissues have distinct metabolic phenotypes, (2) cancer types have drastically different metabolic shifts compared to their adjacent normal controls, and (3) the different shifts happened to tissue-specific metabolic phenotypes result in a converged metabolic phenotype through cancer types and cancer progression. This study strongly suggests the possibility to have a unified framework for studies of cancer-inducing stressors, adaptive metabolic reprogramming, and cancerous behaviors.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.514124v1?rss=1
Authors: Ellis, C. A., Miller, R. L., Calhoun, V. D.
Abstract: Clinicians and developers of deep learning-based neuroimaging clinical decision support systems (CDSS) need to know whether those systems will perform well for specific individuals. However, relatively few methods provide this capability. Identifying neuropsychiatric disorder subtypes for which CDSS may have varying performance could offer a solution. Dynamic functional network connectivity (dFNC) is often used to study disorders and develop neuroimaging classifiers. Unfortunately, few studies have identified neurological disorder subtypes using dFNC. In this study, we present a novel approach with which we identify 5 states of dFNC activity and 4 schizophrenia subtypes based on their time spent in each state. We also show how the performance of an explainable diagnostic deep learning classifier is subtype-dependent. We lastly examine how the dFNC features used by the classifier vary across subtypes. Our study provides a novel approach for subtyping disorders that (1) has implications for future scientific studies and (2) could lead to more reliable CDSS.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.513226v1?rss=1
Authors: Bonilla Yanez, M., Kettlety, S. A., Finley, J., Schweighofer, N., Leech, K. A.
Abstract: Background: Gait biofeedback is commonly used to reduce gait dysfunction in a variety of clinical conditions. In these studies, participants alter their walking to reach the desired magnitude of a specific gait parameter (the biofeedback target) with each step. Biofeedback of parameters such as anterior ground reaction force and step length have been well-studied. Yet, there is no standardized methodology to set the target magnitude of these parameters. Here we present an approach to predict the anterior ground reaction force and step length of neurotypical adults walking at different speeds as a potential method for personalized gait biofeedback. Research question: To determine if anterior ground reaction force and step lengths achieved during neurotypical walking could be predicted using gait speed and participants' demographic and anthropomorphic characteristics. Methods: We analyzed kinetic and kinematic data from 51 neurotypical adults who walked on a treadmill at up to eight speeds. We calculated the average peak anterior ground reaction force and step length of the right lower extremity at each speed. We used linear mixed-effects models to evaluate the effect of speed, leg length, mass, and age on anterior ground reaction force and step length. We fit the model to data from 37 participants and validated predictions from the final models on an independent dataset from 23 participants. Results: Final prediction models for anterior ground reaction force and step length both included speed, speed squared, age, mass, and leg length. The models both showed strong agreement between predicted and actual values on an independent dataset. Significance: Anterior ground reaction force and step length for neurotypical adults can be predicted given an individual's gait speed, age, leg length, and mass. This may provide a standardized method to personalize targets for individuals with gait dysfunction in future studies of gait biofeedback.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.27.513990v1?rss=1
Authors: Watson, S., Porter, H., Sudbery, I., Thompson, R. H.
Abstract: Single cell RNA sequencing (scRNAseq) is a rapidly advancing field which allows for the characterization of the cellular heterogeneity of gene expression profiles within a population. Cell cycle phase is a major contributor to gene expression variance between cells and computational analysis tools have been developed to assign cell cycle phase to scRNAseq datasets. Whilst these tools can be extremely useful, all have the drawback that they classify cells as G1, S or G2/M. Discrete cell phase assignment tool are unable to differentiate between G2 and M and continuous phase assignments tools are unable to identify a region corresponding specifically to mitosis in a pseudo-timeline for continuous assignment along the cell cycle. Bulk RNA sequencing was used to identify differentially expressed genes between mitotic and interphase cells isolated based on phospho-histone H3 expression using fluorescence activated cell sorting. The gene lists were used to develop a Modified Seurat Mitotic Sort (MoSMiS) methodology which can distinguish G2 and M phase cells in single cell RNA sequencing data. The phase assignment tools present in Seurat were modified to allow for cell cycle phase assignment of all stages of the cell cycle identifying a mitotic specific cell population.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513933v1?rss=1
Authors: Ma, R., Sun, E. D., Zou, J.
Abstract: Dimension reduction and data visualization aim to project a high-dimensional dataset to a low-dimensional space while capturing the intrinsic structures in the data. It is an indispensable part of modern data science, and many dimensional reduction and visualization algorithms have been developed. However, different algorithms have their own strengths and weaknesses, making it critically important to evaluate their relative performance for a given dataset, and to leverage and combine their individual strengths. In this paper, we propose an efficient spectral method for assessing and combining multiple visualizations of a given dataset produced by diverse algorithms. The proposed method provides a quantitative measure - the visualization eigenscore - of the relative performance of the visualizations for preserving the structure around each data point. Then it leverages the eigenscores to obtain a consensus visualization, which has much improved quality over the individual visualizations in capturing the underlying true data structure. Our approach is flexible and works as a wrapper around any visualizations. We analyze multiple simulated and real-world datasets from diverse applications to demonstrate the effectiveness of the eigenscores for evaluating visualizations and the superiority of the proposed consensus visualization. Furthermore, we establish rigorous theoretical justification of our method based on a general statistical framework, yielding fundamental principles behind the empirical success of consensus visualization along with practical guidance.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513870v1?rss=1
Authors: Deganutti, G., Pipito, L., Rujan, R. M., Weizmann, T., Griffin, P., Ciancetta, A., Moro, S., Reynolds, C. A.
Abstract: G protein-coupled receptors (GPCRs) are the most abundant membrane proteins and the target of about 35% of approved drugs. Despite this, the structural basis of GPCR pharmacology is still a matter of intense study. Molecular dynamics (MD) simulations aim at expanding our knowledge of GPCR dynamics by building upon the recent advances in structural biology. However, the timescale limitations of classic MD hinder its applicability to numerous structural processes happening in time scales longer than microseconds (hidden structural transitions). For this reason, the overall MD impact on the study of GPCRs pharmacology and drug design is still limited. To overcome this, we have developed an unbiased adaptive sampling algorithm, namely multiple walker supervised MD (mwSuMD), and tested it on different hidden transitions involving GPCRs. By increasing the complexity of the simulated process, we report the binding and unbinding of the vasopressin peptide, the inactive-to-active transition of the glucagon-like peptide-1 receptor (GLP-1R), the stimulatory G protein (Gs) and inhibitory Gi binding to the adrenoreceptor {beta}2 ({beta}2 AR) and the adenosine 1 receptor (A1R) respectively, and the heterodimerization between the adenosine receptor A2 (A2AR) and the dopamine receptor D2 (D2R). We demonstrate that mwSuMD is a helpful tool for studying at the atomic level GPCR transitions that are challenging to address with classic MD simulations.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513936v1?rss=1
Authors: Wedell, E., Shen, C., Warnow, T.
Abstract: Phylogenetic placement, the problem of placing sequences into phylogenetic trees, has been limited either by the number of sequences placed in a single run or by the size of the placement tree. The most accurate scalable phylogenetic placement method with respect to the number of query sequences placed, EPA-ng (Barbera et al., 2019), has a runtime that scales sublinearly to the number of query sequences. However, larger phylogenetic trees cause an increase in the memory usage for EPA-ng, limiting the method to placement trees of up to 10,000 sequences. Our recently designed SCAMPP (Wedell et al., 2021) framework has been shown to scale EPA-ng to larger placement trees of up to 200,000 sequences by building a subtree for the placement of each query sequence. The approach of SCAMPP does not take advantage of the parallel efficiency in EPA-ng since it only places a single query for each run of EPA-ng. Here we present BATCH-SCAMPP, a new technique that overcomes this barrier and enables EPA-ng and other phylogenetic placement methods to scale to ultra-large backbone trees and many query sequences. BATCH-SCAMPP is freely available in GitHub.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513894v1?rss=1
Authors: DeMario, S. M., Xu, K., He, K., Chanfreau, G.
Abstract: RT-PCR and Northern blots have long been used to study RNA isoforms usage for single genes. Recently, advancements in long read sequencing have yielded unprecedented information about the usage and abundance of these RNA isoforms. However, visualization of long-read sequencing data remains challenging due to the high information density. To alleviate these issues we have developed NanoBlot, a simple, open-source, command line tool, which generates Northern blot and RT-PCR-like images from third generation sequencing data. NanoBlot accepts processed bam files. Plotting is based around ggplot2 and is easily customizable. Advantages of NanoBlots include: designing probes to visualize isoforms which would be impossible with traditional RT-PCR or Northern blots, excluding reads from the Nanoblots based on the presence or absence of a specified region and, multiplexing plots with multiple colors. We present examples of NanoBlots compared to actual northern blot data. In addition to traditional gel-like images, NanoBlot also outputs other visualizations such as violin plots. The use of Nanoblot should provide a simple answer to the challenge of visualization of long-read RNA sequencing data.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.512745v1?rss=1
Authors: White, B. S., Woo, X., Koc, S., Sheridan, T., Neuhauser, S. B., Wang, S., Evrard, Y. A., Landua, J. D., Mashl, R. J., Davies, S. R., Fang, B., Raso, M. G., Evans, K. W., Bailey, M. H., Chen, Y., Xiao, M., Rubinstein, J., Foroughi pour, A., Dobrolecki, L. E., Fujita, M., Fujimoto, J., Xiao, G., Fields, R. C., Mudd, J. L., Xu, X., Hollingshead, M. G., Jiwani, S., PDXNet Consortium,, Davis-Dusenbery, B., Wallace, T. A., Moscow, J. A., Doroshow, J. H., Mitsiades, N., Kaochar, S., Pan, C.-x., Chen, M. S., Carvajal-Carmona, L. G., Welm, A. L., Welm, B. E., Govindan, R., Li, S., Davies, M. A., Roth
Abstract: Patient-derived xenografts (PDXs) model human intra-tumoral heterogeneity in the context of the intact tissue of immunocompromised mice. Histological imaging via hematoxylin and eosin (H&E) staining is performed on PDX samples for routine assessment and, in principle, captures the complex interplay between tumor and stromal cells. Deep learning (DL)-based analysis of large human H&E image repositories has extracted inter-cellular and morphological signals correlated with disease phenotype and therapeutic response. Here, we present an extensive, pan-cancer repository of nearly 1,000 PDX and paired human progenitor H&E images. These images, curated from the PDXNet consortium, are associated with genomic and transcriptomic data, clinical metadata, pathological assessment of cell composition, and, in several cases, detailed pathological annotation of tumor, stroma, and necrotic regions. We demonstrate that DL can be applied to these images to classify tumor regions and to predict xenograft-transplant lymphoproliferative disorder, the unintended outgrowth of human lymphocytes at the transplantation site. This repository enables PDX-specific, investigations of cancer biology through histopathological analysis and contributes important model system data that expand on existing human histology repositories. We expect the PDXNet Image Repository to be valuable for controlled digital pathology analysis, both for the evaluation of technical issues such as stain normalization and for development of novel computational methods based on spatial behaviors within cancer tissues.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513827v1?rss=1
Authors: Yuzeir, A., Bejaran, D., Grein, S., Hasenauer, J., Schlitzer, A., Yu, J.
Abstract: To allow the comprehensive histological analysis of the whole intestine in one image, the tissue is often rolled to a spiral before imaging. This Swiss-rolling technique facilitates robust experimental procedures, but it limits the possibilities to comprehend changes along the intestine. Here, we present IntestLine, a Shiny-based open-source application to map imaging data of intestinal tissues in spiral shape onto a line. The mapping of intestinal tissues improves the visualization of the whole intestine in both proximal-distal and serosa-luminal axis, and facilitates the observation of location-specific cell types and markers. In summary, IntestLine serves as a tool to visualize and characterize intestine in future imaging studies.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513890v1?rss=1
Authors: Joudaki, A., Meterez, A., Mustafa, H., Koerkamp, R. G., Kahles, A., Raetsch, G.
Abstract: Sequence-to-graph alignment is an important step in applications such as variant genotyping, read error correction and genome assembly. When a query sequence requires a substantial number of edits to align, approximate alignment tools that follow the seed-and-extend approach require shorter seeds to get any matches. However, in large graphs with high variation, relying on a shorter seed length leads to an exponential increase in spurious matches. We propose a novel seeding approach relying on long inexact matches instead of short exact matches. We demonstrate experimentally that our approach achieves a better time-accuracy trade-off in settings with up to a 25% mutation rate. We achieve this by sketching a subset of graph nodes and storing them in a K-nearest neighbor index. While sketches are more robust to indels, finding the nearest neighbor of a sketch in a high-dimensional space is more computationally challenging than finding exact seeds. We demonstrate that if we store sketch vectors in a K-nearest neighbor index, we can circumvent the curse of dimensionality. Our long sketch-based seed scheme contrasts existing approaches and highlights the important role that tensor sketching can play in bioinformatics applications. Our proposed seeding method and implementation have several advantages: i) We empirically show that our method is efficient and scales to graphs with 1 billion nodes, with time and memory requirements for preprocessing growing linearly with graph size and query time growing quasi-logarithmically with query length. ii) For queries with an edit distance of 25% relative to their length, on the 1 billion node graph, longer sketch-based seeds yield a 4x increase in recall compared to exact seeds. iii) Conceptually, our seeder can be incorporated into other aligners, proposing a novel direction for sequence-to-graph alignment. The implementation is available at: https://github.com/ratschlab/tensor-sketch-alignment.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513897v1?rss=1
Authors: Tabatabaee, Y., Roch, S., Warnow, T.
Abstract: Rooted species trees are used in several downstream applications of phylogenetics. Most species tree estimation methods produce unrooted trees and additional methods are then used to root these unrooted trees. Recently, Quintet Rooting (QR) (Tabatabaee et al., ISMB and Bioinformatics 2022), a polynomial-time method for rooting an unrooted species tree given unrooted gene trees, was introduced. QR, which is based on a proof of identifiability of rooted 5-taxon trees in the presence of incomplete lineage sorting, was shown to have good accuracy, improving over other methods for rooting species trees when incomplete lineage sorting was the only cause of gene tree discordance, except when gene tree estimation error was very high. However, that study left the statistical consistency of QR as an open question. We present QR-STAR, a polynomial-time variant of QR that has an additional step for determining the rooted shape of each quintet tree. We prove that QR-STAR is statistically consistent under the multi-species coalescent (MSC) model. Our simulation study under a variety of model conditions shows that QR-STAR matches or improves on the accuracy of QR. QR-STAR is available in open source form at https://github.com/ytabatabaee/Quintet-Rooting.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.25.513800v1?rss=1
Authors: Ghaffari, S., Bouchonville, K. J., Saleh, E., Schmidt, R. E., Offer, S. M., Sinha, S.
Abstract: Differential gene expression in bulk transcriptomics data can reflect regulated change of transcript abundance within a cell type and/or change in the proportion of cell types within the sample. To differentiate these scenarios, bulk expression deconvolution methods have been developed, which reveal cell type proportions and transcriptomes at the larger scales afforded by bulk RNA-seq compared to single-cell RNA-seq. However, the accuracy of these methods is highly sensitive to technical and biological differences between bulk profiles and the cell type-signatures required as references during deconvolution. We present BEDwARS, a Bayesian deconvolution method specifically designed to address potential differences between reference signatures and true but unknown signatures underlying the bulk transcriptomic profiles. Through extensive benchmarking utilizing eight different datasets derived from pancreas and brain, and by generating additional noisy reference signatures, we demonstrate that BEDwARS outperforms leading in-class methods for estimating cell type proportions and is more robust to noise in reference signatures. Furthermore, it more accurately estimates true cell type signatures compared to the state-of-the-art method. Application of BEDwARS to newly generated RNA-seq and scRNA-seq data on a rare pediatric condition (Dihydropyridine Dehydrogenase deficiency) revealed the possible involvement of ciliopathy and impaired translational control in the etiology of the disorder.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513814v1?rss=1
Authors: Odion, R., Mukdad, L., Alhiyari, Y., Tam, K., Shori, R., Vo-Dinh, T., St. John, M. A.
Abstract: Background: Recent advances in artificial intelligence (AI) in the field of imaging have resulted in new opportunities for automated tumor detection and margin assessment. In particular, AI deep learning techniques such as the Convolutional Neural Network (CNN) have greatly advanced the field of computer vision. Here we introduce the application of a CNN model for use with Dynamic Optical Contrast Imaging (DOCI), an imaging technique developed by our group that creates a unique molecular signature on tissue targets by obtaining the autofluorescence decay of several spectral bands in the UV-Vis range. Methods: 21 patients undergoing surgical resection for tonsillar squamous cell carcinoma (SCC) were identified on a prospective basis. DOCI images were analyzed and compared to the pathology results as ground truth. A CNN model was used to segment sections of DOCI images and provide a percentage chance of tumor presence, allowing for automated tumor margin delineation without a-priori knowledge of the tumor tissue composition. Results: CNN outputs yielded a 99.98% confidence in classifying non-tumor tissue and 76.02% confidence in classifying tumor tissue. Conclusions: Our results indicate that a CNN-based classification model for DOCI allows for real-time analysis of tissue, providing improved sensitivity and accuracy of determining true margins and thus enabling the head and neck cancer surgeon to save healthy tissue and improve patient outcomes.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.25.513785v1?rss=1
Authors: Su, J., Reynier, J.-B., Fu, X., Zhong, G., Jiang, J., Supo Escalante, R., Wang, Y., Izar, B., Knowles, D. A., Rabadan, R.
Abstract: Spatial omics technologies, such as spatial transcriptomics, allow the identification of spatially organized biological processes, while presenting computational challenges for existing analysis approaches that ignore spatial dependencies. Here we introduce Smoother, a unified and modular framework that integrates positional information into non-spatial models via spatial priors and losses. In simulated and real datasets, we show that Smoother enables spatially aware data imputation, cell-type deconvolution, and dimensionality reduction with high accuracy.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513902v1?rss=1
Authors: Bierman, R., Salzman, J.
Abstract: Targeted low-throughput studies have previously identified subcellular RNA localization as necessary for cellular functions including polarization, and translocation. Further, these studies link localization to RNA isoform expression, especially 3 Untranslated Region (UTR) regulation. The recent introduction of genome-wide spatial transcriptomics techniques enable the potential to test if subcellular localization is regulated in situ pervasively. In order to do this, robust statistical measures of subcellular localization and alternative poly-adenylation (APA) at single cell resolution are needed. Developing a new statistical framework called SPRAWL, we detect extensive cell-type specific subcellular RNA localization regulation in the mouse brain and to a lesser extent mouse liver. We integrated SPRAWL with a new approach to measure cell-type specific regulation of alternative 3 UTR processing and detected examples of significant correlations between 3 UTR length and subcellular localization. Included examples, Timp3, Slc32a1, Cxcl14, and Nxph1 have subcellular localization in the brain highly correlated with regulated 3 UTR processing that includes use of unannotated, but highly conserved, 3 ends. Together, SPRAWL provides a statistical framework to integrate multi-omic single-cell resolved measurements of gene-isoform pairs to prioritize an otherwise impossibly large list of candidate functional 3 UTRs for functional prediction and study. SPRAWL predicts 3 UTR regulation of subcellular localization may be more pervasive than currently known.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513955v1?rss=1
Authors: Song, L., Bai, G., Liu, X. S., Li, B., Li, H.
Abstract: Killer immunoglobulin-like receptor (KIR) genes and human leukocyte antigen (HLA) genes are highly polymorphic in a population and play important roles in innate and adaptive immunity. We have developed a novel computational method T1K that can efficiently and accurately infer the KIR or HLA alleles from next-generation sequencing data. T1K is flexible and is compatible with various sequencing platforms including RNA-seq and genomic sequencing data. We applied T1K on CD8+ T cell single-cell RNA-seq data, and identified that KIR2DL4 allele expression levels were enriched in tumor-specific CD8+ T cells.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.25.513701v1?rss=1
Authors: Mitsigkolas, A., Pechlivanis, N., Psomopoulos, F. E.
Abstract: Intro: The number of studies on SARS-CoV-2 published on a daily basis is constantly increasing, in an attempt to understand and address the challenges posed by the pandemic in a better way. Most of these studies also include a phylogeny of SARS-CoV-2 as background context, always taking into consideration the latest data in order to construct an updated tree. However, some of these studies have also revealed the difficulties of inferring a reliable phylogeny. Morel et al., 2021 have shown that reliable phylogeny is an inherently complex task due to the large number of highly similar sequences, given the relatively low number of mutations evident in each sequence. Motivation: From this viewpoint, there is indeed a challenge and an opportunity in identifying the evolutionary history of the SARS-CoV-2 virus, in order to assist the phylogenetic analysis process as well as support researchers in keeping track of the virus and the course of its characteristic mutations, and in finding patterns of the emerging mutations themselves and the interactions between them. The research question is formulated as follows: Detecting new patterns of co-occurring mutations beyond the strain-specific / strain-defining ones, in SARS-CoV-2 data, through the application of ML methods. Aim: Going beyond the traditional phylogenetic approaches, we will be designing and implementing a clustering method that will effectively create a dendrogram of the involved sequences, based on a feature space defined on the present mutations, rather than the entire sequence. Ultimately, this ML method is tested out in sequences retrieved from public databases and validated using the available metadata as labels. The main goal of the project is to design, implement and evaluate a software that will automatically detect and cluster relevant mutations, that could potentially be used to identify trends in emerging variants.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.26.513842v1?rss=1
Authors: Gould, S. I., Sanchez-Rivera, F. J.
Abstract: Prime editing leverages Cas proteins fused to reverse transcription enzymes and prime editing guide RNAs (pegRNAs) to programmably engineer diverse genetic alterations without DNA double strand breaks or exogenous donor templates. Prime editing is powerful, but the process of pegRNA design remains a daunting challenge that limits its widespread adoption by the community. Although a number of computational pegRNA design tools have been developed, none so far contain the right combination of features needed for rapid and systematic design of pegRNAs for variant engineering and high-throughput genetic screens. Here, we describe Prime Editing Guide Generator (PEGG): a computational pipeline for rapid design of pegRNAs and "sensor" pegRNA libraries. PEGG is a user-friendly Python package that generates and visualizes pegRNAs based on a list of input mutations and integrates user-defined properties like variable length of primer binding site and reverse transcription template regions, and outputs ready-to-order pegRNA oligos and libraries containing optimized sequence adapters for pooled cloning. PEGG (including documentation) can be accessed at https://pegg.readthedocs.io.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.25.513682v1?rss=1
Authors: Hackl, S. T., Harbig, T. A., Nieselt, K.
Abstract: The emergence of commercial long read sequencing technologies in the 2010s and the concomitant development of new bioinformatics tools bears the potential of de novo genome assemblies of unprecedented contiguity and quality. However, until today these novel technologies suffer from high rates of sequencing errors. These may be overcome by using long and short reads in combination, in so called hybrid approaches, or by increasing the throughput and thereby the coverage of sequencing runs. In particular the latter will thereby increase the cost of the assembly inevitably. Herein, to-date long read and hybrid assemblers were tested on real whole genome sequencing Illumina and Oxford Nanopore Technologies read data sets and sub samples of these in order to elaborate a best practice for de novo assembly. The findings suggest that although long reads alone can be used to reconstruct complete and contiguous genomes, in particular the single-nucleotide and indel error rate remains high compared to hybrid approaches and that this can impact downstream applications such as variation discovery and gene prediction negatively.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.25.513678v1?rss=1
Authors: Liu, Z., Xu, H., Liu, L., Weng, S., Xing, Z., Ren, Y., Ge, X., Wang, L., Guo, C., Chen, S., Cheng, Q., Luo, P., Zhang, J., Han, X.
Abstract: Abstract Advances in multi-omics and big-data technologies have led to numerous prognostic signatures aimed at improving current clinicopathological staging systems. Due to the lack of reproducibility and independent confirmation, few signatures have been translated into clinical routine. As high-quality datasets accumulate, identifying robust signatures across multiple independent cohorts becomes possible. Nonetheless, inaccurate data retrieval, different versions of genome annotations, disparate expression distributions, difficult data cleaning, inconsistent clinical information, algorithm selection, and parameter tuning have impeded model development and validation in multi-center datasets. Hence, for the first time, we introduced SurvivalML (https://rookieutopia.com/app_direct/SurvivalML/), a web application for helping develop and validate prognostic models across multi-center datasets. SurvivalML included 37,325 samples (253 eligible datasets) with both transcriptome data and survival information from 21 cancer types, which were renewedly and uniformly re-annotated, normalized, and cleaned. This application provided 10 survival machine-learning algorithms for flexibly training models via tuning essential parameters online and delivered four aspects for model evaluation, including Kaplan-Meier survival analysis, time-dependent ROC, calibration curve, and decision curve analysis. Overall, we believe that SurvivalML can serve as an attractive platform for model discovery from multi-center datasets.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.25.513674v1?rss=1
Authors: Soneson, C., Bendel, A. M., Diss, G., Stadler, M. B.
Abstract: Multiplexed assays of variant effect (MAVE) experimentally measure the fitness of large numbers of sequence variants by selective enrichment of sequences with desirable properties followed by quantification by sequencing. mutscan is an R package for flexible analysis of such experiments, covering the entire workflow from raw reads up to statistical analysis and visualization. Core components are implemented in C++ for efficiency. Various experimental designs are supported, including single or paired reads with optional unique molecular identifiers. To find variants with changed relative abundance, mutscan employs established statistical models provided in the edgeR and limma packages. mutscan is available from https://github.com/fmicompbio/mutscan.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.25.513331v1?rss=1
Authors: Gudbjartsson, H., Isleifsson, H. b., Ragnarsson, B., Guimaraes, R., Wu, H., Olafsdottir, H., Stefasson, S. K.
Abstract: Motivation: Our aim was to simplify and speedup joint-genotyping, from sequence based variation data of individual samples, while maintaining as high sensitivity and specificity as possible. Results: We have leveraged versatile GOR data structures to store biallelic representations of variants and sequence read coverage in a very efficient way, allowing for very fast joint-genotyping that is an order of magnitude faster than any joint-genotyping method published to date. Furthermore, it can be easily extended and executed much faster in an incremental fashion. Concordance analysis based on the Genome In A Bottle (GIAB) samples shows favorable results when compared with the de-facto standard approach, using gVCF files and GATK joint-calling. Additionally, we have developed variant quality classification using XGBoost and variant training sets derived from the GIAB samples. The entire business logic is implemented efficiently and concisely in SparkGOR.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513504v1?rss=1
Authors: Valeriani, L., Cuturello, F., Ansuini, A., Cazzaniga, A.
Abstract: Protein language models (pLMs) transform their input into a sequence of hidden representations whose geometric behavior changes across layers. Looking at fundamental geometric properties such as the intrinsic dimension and the neighbor composition of these representations, we observe that these changes highlight a pattern characterized by three distinct phases. This phenomenon emerges across many models trained on diverse datasets, thus revealing a general computational strategy learned by pLMs to reconstruct missing parts of the data. These analyses show the existence of low-dimensional maps that encode evolutionary and biological properties such as remote homology and structural information. Our geometric approach sets the foundations for future systematic attempts to understand the space of protein sequences with representation learning techniques.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513593v1?rss=1
Authors: Klie, A., Stites, H., Jores, T., Carter, H.
Abstract: Deep learning (DL) has become a popular tool to study cis-regulatory element function. Yet efforts to design software for DL analyses in genomics that are Findable, Accessible, Interoperable and Reusable (FAIR) have fallen short of fully meeting these criteria. Here we present EUGENe (Elucidating the Utility of Genomic Elements with Neural Nets), a FAIR toolkit for the analysis of labeled sets of nucleotide sequences with DL. EUGENe consists of a set of modules that empower users to execute the key functionality of a DL workflow: 1) extracting, transforming and loading sequence data from many common file formats, 2) instantiating, initializing and training diverse model architectures, and 3) evaluating and interpreting model behavior. We designed EUGENe to be simple; users can develop and deploy workflows on new or existing datasets with functions that act on two customizable Python objects, annotated sequence data (SeqData) and PyTorch models (BaseModel). The modularity and simplicity of EUGENe also make it highly extensible and we illustrate these principles through application of the toolkit to three predictive modeling tasks. First, we train and compare a set of built-in models along with a custom architecture for the accurate prediction of activities of plant promoters from STARR-seq data. Next, we apply EUGENe to an RNA binding prediction task and showcase how seminal model architectures can be retrained in EUGENe or imported from Kipoi. Finally, we train models to classify transcription factor binding by wrapping functionality from Janngu, which can efficiently extract sequences in BED file format from the human genome. We emphasize that the code used in each use case is simple, readable, and well documented (https://eugene-tools.readthedocs.io/en/latest/index.html). We believe that EUGENe represents a springboard toward a collaborative ecosystem for DL applications in genomics research. EUGENe is available for download on GitHub (https://github.com/cartercompbio/EUGENe) along with several introductory tutorials and for installation on PyPi (https://pypi.org/project/eugene-tools/).
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513546v1?rss=1
Authors: Sun, Z., Liu, J., Zhang, M., Wang, T., Huang, S., Weiss, S. T., Liu, Y.-Y.
Abstract: Accurate species identification and abundance estimation are critical for the interpretation of whole metagenome shotgun sequencing (WMS) data. Numerous computational methods, broadly referred to as metagenomic profilers, have been developed to identify species in microbiome samples by classification of sequencing reads and quantification of their relative abundances. Yet, existing metagenomic profilers typically suffer from false positive identifications and consequently biased relative abundance estimation (as false positives can be accounted for more than 90% of total identified species). Here, we present a new metagenomic profiler MAP2B (MetAgenomic Profiler based on type IIB restriction site) to resolve those issues. We first illustrate the pitfalls of using relative abundance as the only feature in determining false positives. We then propose a feature set to distinguish false positives from true positives. By benchmarking the performance in metagenomic profiling using data from CAMI2 (Critical Assessment of Metagenome Interpretation: second round of challenge), we illustrate the superior performance of MAP2B (F1 score ~ 0.93) over existing metagenomic profilers (F1 score ranges from 0.18 to 0.58). We further tested the performance of MAP2B using real WMS data from an ATCC mock community, confirming its superior performance and robustness against sequencing depth. In addition, by leveraging WMS data from an IBD cohort, we demonstrate the taxonomic features obtained by MAP2B can better discriminate disease status and predict metabolomic profiles.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513564v1?rss=1
Authors: Ethan, C., Maxime, C., Claudio, A. F., Francisca, F. V., Franck, P., Elaine, D. N., Guillaume, B., Auguste, G.
Abstract: Drug discovery uses high throughput screening to identify compounds that interact with a molecular target or that alter a phenotype favorably. The cautious selection of molecules used for such a screening is instrumental, and is tightly related to the hit rate. In this work, we wondered if Cell Painting, a general purpose image-based assay, could be used as an efficient proxy for compound selection, thus increasing the success rate of a specific assay. To this end, we considered Cell Painting images with 30,000 molecules treatments, and selected compounds that produced a visual effect close to the positive control of an assay, by using the Frechet Inception Distance. We then compared the hit rates of such a preselection with what was actually obtained in real screening campaigns. As a result, Cell Painting would have permitted a significant increase in the success rate and, even for one of the assays, would have allowed to reach 80% of the hits with ten times fewer compounds to test. We conclude that images of a Cell Painting assay can be directly used for compound selection prior to screening, and we provide a simple quantitative approach in order to do so.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513557v1?rss=1
Authors: Henfrey, C., Murphy, S., Tellier, M.
Abstract: Transcription and co-transcriptional processes, including pre-mRNA splicing and mRNA cleavage and polyadenylation, regulate the production of a mature mRNA. The carboxyl terminal domain (CTD) of RNA polymerase (pol) II, which comprises 52 repeats of the Tyr1Ser2Pro3Thr4Ser5Pro6Ser7 peptide, is involved in the coordination of transcription with co-transcriptional processes. The pol II CTD is dynamically modified by protein phosphorylation, which regulates recruitment of transcription and co-transcriptional factors. We have investigated whether cytoplasmic levels of mature mRNA from intron-containing protein-coding genes can be inferred from RNA processing efficiency, CTD phosphorylation, and/or association of proteins complexes regulating RNA production. Surprisingly, we found that genes associated with efficient RNA processing have relatively low phosphorylation of the pol II CTD. In contrast, protein-coding genes that produce a low level of mature mRNA are associated with high pol II CTD phosphorylation, poor RNA processing, and increased chromatin retention. Unexpectedly, the transcripts from this subset of genes are not actively degraded by the RNA exosome. RNA exosome-regulated genes are instead characterised by Tyr1 hyperphosphorylation and Integrator recruitment around the poly(A) site. Our results indicate that in addition to the transcription level, CTD phosphorylation and RNA processing efficiency play important roles in the regulation of chromatin retention of transcripts.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513533v1?rss=1
Authors: Drost, F. R., Schiefelbein, L., Schubert, B.
Abstract: T cell receptors (TCRs) bind to pathogen- or self-derived epitopes to elicit a T cell response as part of the adaptive immune system. Determining the specificity of TCRs provides context for immunological studies and can be used to identify candidates for novel immunotherapies. To avoid costly experiments, large-scale TCR-epitope databases are queried for similar sequences via various distance functions. Here, we developed the deep-learning based distance meTCRs. Contrary to most previous approaches, the method avoids computational expansive pairwise string operations by comparing TCRs in a numeric embedding. In contrast to models which are trained specificity-agnostic, we directly utilize epitope information by applying deep metric learning to guide the training. Summarizing, we present meTCRs as a scalable alternative to embed TCR repertoires for clustering, visualisation, and querying against the ever-increasing amount TCR-epitope pairs in publicly available databases.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513545v1?rss=1
Authors: Wang, B., Van Der Kloet, F., Kes, M., Luirink, J., Hamoen, L. W.
Abstract: When analysing transcriptome data, threshold values are chosen to decide whether the regulation of a gene is relevant or not, however this may result in the loss of valuable information. To overcome this problem, it can be useful to analyse regulons instead of individual genes, to harness the statistical power of combining genes. Another advantage of a regulon-based analysis is that it provides direct insights into the activity of regulatory pathways, which is the essence of transcriptome analyses. We have developed a software tool called GINtool that can use regulon information to analyse transcriptome data. GINtool includes the option to take the activity mode of a regulator in account, which is important when a regulator can function both as an activator and repressor. GINtool also contains two novel graphical representations that greatly facilitate the visual inspection of regulon-based transcriptome analyses. Additional features of GINtool includes the evaluation of transcriptome data using functional categories, and the analysis of gene expression differences within operons. To ease the analyses and downstream processing of figures, GINtool has been developed as an add-in for Excel.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513543v1?rss=1
Authors: Cao, S., He, C., Zhao, X., Yu, R., Li, Y., Fang, W., Zhang, C.-Y., Yan, W., Chen, D.
Abstract: Plant morphogenesis relies on precise gene expression programs at the proper time and position which is orchestrated by transcription factors (TFs) in intricate regulatory networks at a cell-type specific manner. Here we presented a reference single-cell transcriptomic atlas of Arabidopsis seedlings by integration of 40 published scRNA-seq datasets from representative tissues as well as the entire under- and above-ground parts. We identified 34 distinct cell types or states, largely expanding our current view of plant cell compositions. We then mapped the developmental trajectory of root-shoot lineage separation and identified differential gene expression programs that may regulate the cell fate determination of under- and above-ground organs. Lastly, we systematically constructed cell-type specific gene regulatory networks and uncovered key regulators that act in a coordination manner to control cell-type specific gene expression. Taken together, our study not only offers a valuable resource plant cell atlas exploration but also provides molecular insights into gene-regulatory programs that determines organ specify, particularly the differentiation between root and shoot.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513610v1?rss=1
Authors: Kim, D. G., Kim, H. S., Choi, Y., Stan, R. C.
Abstract: Fever is a typical symptom of most infectious diseases. While prolonged fever may be clinically undesirable, mild reversible fever ( less than 39, 312K) can potentiate the immune responses against pathogens. Here, using molecular dynamics, we investigated the effect of febrile temperatures (38 to 40, 311K to 313K) on the immune complexes formed by the SARS-CoV-2 spike protein with two neutralizing antibodies. We found that, at mild fever temperatures (311-312K), the binding affinities of the two antibodies improve when compared to the physiological body temperature (37, 310K). Furthermore, only at 312K, antibodies exert distinct mechanical effects on the receptor binding domains of the spike protein that may hinder SARS-CoV-2 infectivity. Enhanced antibody binding affinity may thus be obtained using appropriate temperature conditions.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513521v1?rss=1
Authors: Manfredi, M., Savojardo, C., Martelli, P. L., Casadio, R.
Abstract: The knowledge of protein-protein interaction sites (PPIs) is crucial for protein functional annotation. Here we address the problem focusing on the prediction of putative PPIs having as input protein sequences. The problem is important given the huge volume of sequences compared to experimental and/or computed protein structures. Taking advantage of recently developed protein language models and Deep Neural networks here we describe ISPRED-SEQ, which overpasses state-of-the-art predictors addressing the same problem. ISPRED-SEQ is freely available for testing at https://ispredws.biocomp.unibo.it.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513552v1?rss=1
Authors: Pillich, R. T., Chen, J., Churas, C., Fong, D., Gyori, B. M., Ideker, T., Karis, K., Liu, S. N., Ono, K., Pico, A., Pratt, D.
Abstract: The investigation of sets of genes using biological pathways is a common task for researchers and is supported by a wide variety of software tools. This type of analysis generates hypotheses about the biological processes active or modulated in a specific experimental context. The NDEx Integrated Query (IQuery) is a new tool for network and pathway-based gene set interpretation that complements or extends existing resources. It combines novel sources of pathways, integration with Cytoscape, and the ability to store and share analysis results. The IQuery web application performs multiple gene set analyses based on diverse pathways and networks stored in NDEx. These include curated pathways from WikiPathways and SIGNOR, published pathway figures from the last 27 years, machine-assembled networks using the INDRA system, and the new NCI-PID v2.0, an updated version of the popular NCI Pathway Interaction Database. IQuery's integration with MSigDB and cBioPortal now provides pathway analysis in the context of these two resources. IQuery is available at https://www.ndexbio.org/iquery and is implemented in Javascript and Java.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513506v1?rss=1
Authors: Feng, C., Wang, W., Han, R., Wang, Z., Ye, L., Du, Z., Wei, H., Zhang, F., Peng, Z., Yang, J.
Abstract: RNA 3D structure prediction remains challenging though after years of efforts. Inspired by the recent breakthrough in protein structure prediction, we developed trRosettaRNA, a novel deep learning-based approach to de novo prediction of RNA 3D structure. Like trRosetta, the trRosettaRNA pipeline comprises two major steps: 1D and 2D geometries prediction by a transformer network; and full-atom 3D structure folding by energy minimization with constraints from the predicted geometries. We benchmarked trRosettaRNA on two independent datasets. The results show that trRosettaRNA outperforms other conventional methods by a large margin. For example, on 25 targets from the RNA-Puzzles experiments, the mean RMSD of the models predicted by trRosettaRNA is 5.5 [A], compared with 10.5 [A] from the state-of-the-art human group (i.e., Das). Further comparisons with two recently released deep learning-based methods (i.e., DeepFoldRNA and RoseTTAFoldNA) show that all three methods have similar accuracy. However, trRosettaRNA yields more accurate and physically more realistic side-chain atoms than DeepFoldRNA and RoseTTAFoldNA. Finally, we apply trRosettaRNA to predict the structures for the Rfam families that do not have known structures. Analysis shows that for 263 families, the predicted structure models are estimated to be accurate with RMSD less than 4 [A]. The trRosettaRNA server and the package are available at: https://yanglab.nankai.edu.cn/trRosettaRNA/.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.512549v1?rss=1
Authors: Lippeveld, M., Peralta, D., Filby, A., Saeys, Y.
Abstract: Morphologically profiling large-scale, single-cell bioimaging datasets poses a significant computational challenge. Here, we present Scalable Cytometry Image Processing (SCIP), a software package implemented in Python and aimed at running on high performance computing infrastructure. SCIP is scalable, flexible, open-source and enables reproducible image processing. It performs projection, illumination correction, segmentation, background masking and extensive feature extraction. We showcase SCIP's capabilities on three large, bioimaging datasets. First, we process an imaging flow cytometry (IFC) dataset of human white blood cells and show how the obtained features are used to classify the cells into 8 cell types based on bright- and darkfield imagery. Secondly, we process an automated microscopy dataset of human white blood cells to divide them into cell types in an unsupervised manner. Finally, a high-content screening dataset of breast cancer cells is processed to predict the mechanism-of-action of a large set of compound treatments. The software is available for install from the PyPi repository. The source code is available at https://github.com/ScalableCytometryImageProcessing/SCIP under the GNU General Public License version 3. It has been tested on Unix operating systems. Issues with the software can be submitted at https://github.com/ScalableCytometryImageProcessing/SCIP/issues.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513465v1?rss=1
Authors: Carbone, G., Cuturello, F., Bortolussi, L., Cazzaniga, A.
Abstract: Deep Learning models for protein structure prediction, such as AlphaFold2, leverage Transformer architectures and their attention mechanism to capture structural and functional properties of amino acid sequences. Despite the high accuracy of predictions, biologically insignificant perturbations of the input sequences, or even single point mutations, can lead to substantially different 3d structures. On the other hand, protein language models are often insensitive to biologically relevant mutations that induce misfolding or dysfunction (e.g. missense mutations). Precisely, predictions of the 3d coordinates do not reveal the structure-disruptive effect of these mutations. Therefore, there is an evident inconsistency between the biological importance of mutations and the resulting change in structural prediction. Inspired by this problem, we introduce the concept of adversarial perturbation of protein sequences in continuous embedding spaces of protein language models. Our method relies on attention scores to detect the most vulnerable amino acid positions in the input sequences. Adversarial mutations are biologically diverse from their references and are able to significantly alter the resulting 3d structures.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513505v1?rss=1
Authors: Braun, J., Neme, R., Feng, Y., Landweber, L. F., Jonoska, N.
Abstract: DNA rearrangements are important in various contexts, such as in vertebrate immunity, and cancer genome instability. The single-celled eukaryote Oxytricha trifallax undergoes massive and reproducible genome rearrangement during post-zygotic development, making it a compelling model organism to study DNA rearrangements. To date, computational tools for the extraction and analysis of rearrangement annotations lack transparency and rely on assumptions that may not hold for all analyzed data, leading to irreproducibility of results and loss of information through data filtering or misrepresentation. An implementation of a procedure for the annotation and analysis of DNA rearrangement as a web application is discussed and tested. The resulting annotations provide an improvement over previous annotations in the following manner. (a) SDRAP achieves more complete precursor-product mappings than previous software (b) the software allows for full transparency of all parameters used during the annotation and therefore facilitates reproducible results, and (c) this parameter transparency makes SDRAP suitable for comparison of genomic data from different sources, including cancer genomes. This work introduces a theoretical framework and software to systematically extract and analyze annotations from pairs of genome assemblies corresponding to precursor and product rearrangement data. The software makes no assumptions about the structure of the rearrangements, and permits the user to select parameters to suit the data.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513498v1?rss=1
Authors: Senanayake, A., Gamaarachchi, H., Herath, D., Ragel, R.
Abstract: Nanopore sequencing allows selective sequencing, the ability to programmatically reject unwanted reads in a sample. Selective sequencing has many present and future applications in genomics research and the classification of species from a pool of species is an example. Existing methods for selective sequencing for species classification are still immature and the accuracy highly varies depending on the datasets. For the five datasets we tested, the accuracy of existing methods varied in the range of ~77%-97% (average accuracy less than 89%). Here we present DeepSelectNet, an accurate deep-learning-based method that can directly classify nanopore current signals belonging to a particular species. DeepSelectNet utilizes novel data preprocessing techniques and improved neural network architecture for regularization. For the five datasets tested, DeepSelectNet's accuracy varied between ~91%-99% (average accuracy ~95%). At its best performance, DeepSelectNet achieved a nearly 12% accuracy increase compared to its deep learning-based predecessor SquiggleNet. Furthermore, precision and recall evaluated for DeepSelectNet on average were always greater than 89% (average ~95%). In terms of execution performance, DeepSelectNet outperformed SquiggleNet by ~13% on average. Thus, DeepSelectNet is a practically viable method to improve the effectiveness of selective sequencing. Compared to base alignment and deep learning predecessors, DeepSelectNet can significantly improve the accuracy to enable real-time species classification using selective sequencing. The source code of DeepSelectNet is available at https://github.com/AnjanaSenanayake/DeepSelectNet.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513174v1?rss=1
Authors: Cracco, A., Tomescu, A. I.
Abstract: Compacted de Bruijn graphs are one of the most fundamental data structures in computational genomics. Colored compacted graphs Bruijn graphs are a variant built on a collection of sequences, and associate to each k-mer the sequences in which it appears. Here we present GGCAT, a tool for constructing both types of graphs. Compared to Cuttlefish 2 (Genome Biology, 2022), the state-of-the-art for constructing compacted de Bruijn graphs, GGCAT has a speedup of up to 3.4x for k = 63 and up to 20.8x for k = 255. Compared to Bifrost (Genome Biology,2020), the state-of-the-art for constructing the colored variant, GGCAT achieves a speedup of up to 12.6x for k = 27. GGCAT is up to 480x faster than BiFrost for batch sequence queries on colored graphs. GGCAT is based on a new approach merging the k-mer counting step with the unitig construction step, and on many practical optimizations. GGCAT is implemented in Rust and is freely available at https://github.com/algbio/ggcat This work was partially funded by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement No. 851093, SAFEBIO), and partially by the Academy of Finland (grants No. 322595, 352821, 346968).
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513494v1?rss=1
Authors: Shetab Boushehri, S., Essig, K., Chlis, N.-K., Herter, S., Bacac, M., Theis, F. J., Glasmacher, E., Marr, C., Schmich, F.
Abstract: Therapeutic antibodies are widely used to treat severe diseases. Most of them alter immune cells and act within the immunological synapse; an essential cell-to-cell interaction to direct the humoral immune response. Although many antibody designs are generated and evaluated, a high-throughput tool for systematic antibody characterization and prediction of function is lacking. Here, we introduce the first comprehensive open-source framework, scifAI (single-cell imaging flow cytometry AI), for preprocessing, feature engineering and explainable, predictive machine learning on imaging flow cytometry (IFC) data. Additionally, we generate the largest publicly available IFC data set of the human immunological synapse containing over 2.8 million images. Using scifAI, we analyze class frequency- and morphological changes under different immune stimulation. T cell cytokine production across multiple donors and therapeutic antibodies is quantitatively predicted in vitro, linking morphological features with function and demonstrating the potential to significantly impact antibody design. scifAI is universally applicable to IFC data, and, given its modular architecture, straightforward to incorporate into existing workflows and analysis pipelines, e.g. for rapid antibody screening and functional characterization.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513476v1?rss=1
Authors: Pellechia, S., Viscido, G., Franchini, M., Gambardella, G.
Abstract: Although an essential step, the functional annotation of cells often proves particularly challenging in the analysis of single-cell transcriptional data. Several methods have been developed to accomplish this task. However, in most cases, these rely on techniques initially developed for bulk RNA sequencing or simply make use of marker genes identified from cell clustering followed by supervised annotation. To overcome these limitations and automatise the process, we have developed two novel methods, the single-cell gene set enrichment analysis (scGSEA) and the single cell mapper (scMAP). scGSEA combines latent data representations and gene set enrichment scores to detect coordinated gene activity at single-cell resolution. scMAP uses transfer learning techniques to re-purpose and contextualise new cells into a reference cell atlas. Using both simulated and real datasets, we show that scGSEA effectively recapitulates recurrent patterns of pathways' activity shared by cells from different experimental conditions. At the same time, we show that scMAP can reliably map and contextualise new single cell profiles on a breast cancer atlas we recently released. Both tools are provided in an effective and straightforward workflow providing a framework to determine cell function and significantly improve annotation and interpretation of scRNA-seq data.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513446v1?rss=1
Authors: Yang, S., Ha, K., Song, W., Fujita, M., K?bler, K., Polak, P., de Alba Rivas, C. G., Pessina, P., de Aja, J. S., Rowbotham, S., Bhetariya, P., Dost, A., Moye, A. L., Hiyama, E., Nakagawa, H., Kim, C. F., Kim, H.-G., Lee, H.
Abstract: We here present COOBoostR (https://github.com/SWJ9385/COOBoostR), a computational method designed for the putative prediction of tissue- or cell-of-origin of various cancer types. COOBoostR leverages regional somatic mutation density information and chromatin mark features to be applied to an extreme gradient boosting-based machine-learning algorithm. COOBoostR ranks chromatin marks from various tissue and cell types which best explain the somatic mutation density landscape of any sample of interest. Through integrating either ChIP-seq based chromatin data or bulk/single cell chromatin accessibility data along with regional somatic mutation density data derived from normal cells/tissue, precancerous lesions, and cancer types, we show that COOBoostR outperforms existing random forest-based methods in prediction speed with comparable or better tissue or cell-of-origin prediction performance. In addition, our results suggest a dynamic somatic mutation accumulation at the normal tissue or cell stage which could be intertwined with the changes in open chromatin marks and enhancer sites. These results further represent chromatin marks shaping the somatic mutation landscape at the early stage of mutation accumulation, possibly even before the initiation of precancerous lesions or neoplasia.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513539v1?rss=1
Authors: Castanza, A. S., Recla, J. M., Eby, D., Thorvaldsdottir, H., Bult, C. J., Mesirov, J. P.
Abstract: The Molecular Signatures Database (MSigDB) serves as the primary repository of biological signature gene sets for performing Gene Set Enrichment Analysis (GSEA). In the more than 15 years since its creation, MSigDB has served over 290,000 users in their use of GSEA to perform statistically rigorous analysis of coordinated patterns of gene expression changes by leveraging the prior knowledge of tens of thousands of deposited signatures. In that time, the sets provided in MSigDB have been offered exclusively in the human gene space and only minimally supporting analysis of mouse model data through mapping to human genes. Here we present two substantial improvements to MSigDB: first, by providing gene sets from widely used resources in the mouse gene space; and second, by offering substantially improved orthology mapping resources for comparative analysis of both mouse and human datasets.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513588v1?rss=1
Authors: Carpanzano, S., Santorsola, M., nf-core community,, Lescai, F.
Abstract: Horizontal gene transfer (HGT) is well described in prokaryotes, it plays a crucial role in evolution, and has functional consequences in insects and plants: less is known about HGT in Humans. Studies have reported bacterial integrations in cancer patients, and microbial sequences have been detected in data from well-known Human sequencing projects. Few of the existing tools to investigate HGT are highly automated. Thanks to the adoption of Nextflow for life sciences workflows, and the standards and best practices curated by communities such as nf-core, fully automated, portable, and scalable pipelines can now be developed. Here we present nf-core/hgtseq, to facilitate the analysis of HGT from sequencing data in different organisms. We showcase its performance by analysing six exome datasets from five mammals. Hgtseq can be run seamlessly in any computing environment and accepts data generated by existing exome and whole-genome sequencing projects: this will enable researchers to expand their analyses into this area. Fundamental questions are still open, about the mechanisms and the extent or the role of horizontal gene transfer: by releasing hgtseq we provide a standardised tool which will enable a systematic investigation of this phenomenon, thus paving the way for a better understanding of HGT.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.20.513118v1?rss=1
Authors: Yang, Z., Ye, Z., Qiu, J., Feng, R., Li, D., Hsieh, C., Allcock, J., Zhang, S.
Abstract: Mutation-induced drug resistance -- where the efficacy of drugs is diminished by structural changes in proteins -- presents a significant challenge to drug development and the clinical treatment of disease. Understanding the effects of mutation on protein-ligand binding affinities is a key step in developing more effective drugs and therapies, but as a research community we are currently hindered by the lack of a comprehensive database of relevant information. To address this issue, we have developed MdrDB, a database of information related to changes in protein-ligand affinity caused by mutations in protein structure. MdrDB combines data from seven publicly available datasets with calculated biochemical features, as well as 3D structures computed with PyMOL and AlphaFold 2.0, to form the largest database of its kind. With 3D structural information provided for all samples, MdrDB was specifically created to have the size, breadth, and complexity to be useful for practical protein mutation studies and drug resistance modeling. The database brings together wild type and mutant protein-ligand complexes, binding affinity changes upon mutation ($Delta Delta$G), and biochemical features calculated from complexes to advance our understanding of mutation-induced drug resistance, the development of combination therapies, and the discovery of novel chemicals. In total, MdrDB contains 100,537 samples generated from 240 proteins (5,119 total PDB structures), 2,503 mutations, and 440 drugs. Of the total samples, 95,971 are based on available PDB structures, with the remaining 4,566 based on AlphaFold 2.0 predicted structures.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.23.513389v1?rss=1
Authors: Zhao, B., Wei, D., Xiong, Y., Ding, J.
Abstract: The ever-increasing availability of single-cell transcriptomic data offers unrivaled opportunities to profile cellular states in various biological processes at high resolution, which has brought substantial advancements in understanding complex mechanisms underlying a large variety of bioprocesses. As limited by the protocol and technology, single-cell measurements in one study are often performed in batches, which unavoidably induces biological and technical differences in the single-cell measurements of the same study. Consequently, it presents challenges in analyzing all single-cell data from different batches together, particularly if the measurements were assayed with different technologies. Several methods have been developed to remove the aforementioned batch effects recently. However, there remain challenges unaddressed with those existing methods, including but not limited to the risk of over-correction, the need for the assumption of gene expression distribution, and expensive computation. To mitigate those limitations, we develop a novel deep learning method called scCobra that combines contrastive learning, domain adaptation, and generative adversarial networks to remove batch effects in single-cell RNA-seq data. The contrastive learning network is utilized to learn latent embeddings to represent the cells, domain-adaptation is employed to batch-normalize the latent embeddings of cells from distinct batches, while generative adversarial networks further optimize the blending effect. The proposed method does not require any prior assumption of gene expression distribution. We applied the scCobra method to one simulated and two real single-cell datasets with significantly experimental differences. Our method outperforms other benchmarked methods in batch correction and biological conservation, and its running efficiency is also among the best.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.22.513328v1?rss=1
Authors: Arnaudi, M., Beltrame, L., Degn, K., Utichi, M., Scrima, S., Sackett, P. W., Lambrughi, M., Tiberti, M., Papaleo, E.
Abstract: Cancer is a complex group of diseases due to the accumulation of mutations in tumor suppressors or oncogenes in the genome. Cancer alterations can be very heterogeneous, even in tumors from the same tissue, affecting the response to treatment and risks of relapse in different patients. The role of genomics variants on cancer predisposition, progression, and response to treatment continues to be realized. Thanks to advances in sequencing techniques and their introduction in a clinical setting, the number of genomic variants discovered is growing exponentially. Many of these variants are classified as Variants of Uncertain Significance (VUS), while other variants have been reported with conflicting evidence. Applications of bioinformatic-based approaches to characterize the effects of these variants demonstrated their full potential thanks to advances in machine learning, comparisons between predicted effects and cellular readouts, and advances in the field of structural biology and biomolecular simulations. We here introduce a modular structure-based framework for the annotations and classification of the impact of variants affecting the coding region of genes and impacting on the corresponding protein product (MAVISp, Multi-layered Assessment of VarIants by Structure for proteins) together with a Streamlit-based web application (https://github.com/ELELAB/MAVISp) where the variants and the data generated by the assessment are made available to the community for consultation or further studies. Currently, MAVISp includes information for ten different proteins and more than 4000 variants. New protein targets are routinely analyzed in batches through standardized Python-based workflows and high-throughput free energy and biomolecular simulations. We also illustrate the potential of the approach for each protein included in the database. New variants will be deposited on a regular base or in connection with future publications where the approach will be applied. Finally, we provide guidelines for new contributors who are interested in contributing to the collection in relation to their research.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513123v1?rss=1
Authors: Cui, R., Elzur, R. A., Kanai, M., Ulirsch, J. C., Weissbrod, O., Daly, M., Neale, B., Fan, Z., Finucane, H. K.
Abstract: Fine-mapping aims to identify genetic variants that causally impact a given phenotype. State-of-the-art Bayesian fine-mapping algorithms (for example: SuSiE1, FINEMAP2,3, ABF4, and COJO5-ABF) are widely applied in practice6-11, but it remains challenging to assess their calibration (i.e., whether or not the posterior probability of causality reflects the true proportion of causal variants) in real data, where model misspecification almost certainly exists and true causal variants are unknown. Here, we present the Replication Failure Rate (RFR), a metric to assess the consistency of fine-mapping results based on downsampling a large cohort. Empirical evaluation of fine-mapping results from SuSiE, FINEMAP and COJO-ABF suggest that these methods may be miscalibrated in the under-conservative direction. Next, we show in simulations that non-sparse genetic architecture can lead to miscalibration, while imputation noise, non-normal effect size distributions, and quality control filters removing potentially causal variants are less likely contributors. Here, we present two new fine-mapping methods, SuSiE-inf and FINEMAP-inf, that extend SuSiE and FINEMAP to incorporate a term for infinitesimal effects in addition to a small number of larger causal effects of interest. Our methods exhibit better calibration in simulations and improved RFR and functional enrichment in real data, with minimal loss of recall and competitive computational cost. Furthermore, using the sparse fine-mapped variants identified by our methods to perform cross-population genetic risk prediction in the UK Biobank, we observed a substantial increase in predictive accuracy over SuSiE and FINEMAP. Our work improves our ability to pinpoint causal variants for complex traits, a fundamental goal of human genetics.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.22.513294v1?rss=1
Authors: Ko, S., Chu, B. B., Peterson, D., Okenwa, C., Papp, J. C., Alexander, D. H., Sobel, E. M., Zhou, H., Lange, K. L.
Abstract: Admixture estimation plays a crucial role in ancestry inference and genomewide association studies (GWAS). Computer programs such as ADMIXTURE and STRUCTURE are commonly employed to estimate the admixture proportions of sample individuals. However, these programs can be overwhelmed by the computational burdens imposed by the 10^5 to 10^6 samples and millions of markers commonly found in modern biobanks. An attractive strategy is to run these programs on a set of ancestry informative SNP markers (AIMs) that exhibit substantially different frequencies across populations. Unfortunately, existing methods for identifying AIMs require knowing ancestry labels for a subset of the sample. This supervised learning approach creates a chicken and the egg scenario. In this paper, we present an unsupervised, scalable framework that seamlessly carries out AIM selection and likelihood-based estimation of admixture proportions. Our simulated and real data examples show that this approach is scalable to modern biobank data sets. Our implementation of the method is called OpenADMIXTURE.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.22.513362v1?rss=1
Authors: Chen, E., Liao, R., Shalaginov, M. Y., Zeng, T. H.
Abstract: Acute lymphoblastic leukemia (ALL) is one of the most common types of cancer among children. It can rapidly become fatal within weeks, hence early diagnosis is critical. Problematically, the ALL diagnosis mainly involves manual blood smear analysis relying on the expertise of medical professionals, which is error-prone and time-consuming. Thus, it is important to develop artificial intelligence tools that will identify leukemic cells from a microscopic image faster, more accurately, and cheaper. Here, we investigate the capabilities of a traditional convolutional neural network (CNN) and You Only Look Once (YOLO) models for real-time detection of leukemic cells. The YOLOv5s model shows 97.2% accuracy for the task of object detection of ALL cells, with the inference speed allowing 80 image frames to be processed per second. These new findings can provide valuable insight in applying real-time object detection algorithms for improving the efficiency of blood cancer diagnosis.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.22.512683v1?rss=1
Authors: Barradas-Bautista, D., Almajed, A., Oliva, R., Cavallo, L., Kalnis, P.
Abstract: Protein-protein interactions drive many important biological events, such as infection, replication, and recognition. We need to access the molecular details of the interaction provided by experimental 3D structures to control or engineer such events. However, such experiments take time and are expensive; moreover, the current technology cannot keep up with the high discovery rate of new interactions. Computational modeling like protein-protein docking can help to fill this gap by generating docking poses. Protein-protein docking generally consists of two parts, sampling and scoring. The sampling is an exhaustive search of the tridimensional space. The caveat of the sampling produces a large number of incorrect poses, producing a highly unbalanced dataset. This limits the utility of the data to train machine learning classifiers. Using weak supervision, we developed a data augmentation method named hAIkal. Using hAIkal, we increased the labeled training data to train several algorithms. We trained and obtained different classifiers; the best classifier has 81% accuracy and 0.51 MCC on the test set, surpassing the state-of-the-art scoring functions.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.23.513417v1?rss=1
Authors: Salz, R., Saraiva-Agostinho, N., Vorsteveld, E., van der Made, C. I., Kersten, S., Stemerdink, M., Allen, J., Volders, P.-J., Hunt, S. E., Hoischen, A., 't Hoen, P. A. C.
Abstract: Our incomplete knowledge of the human transcriptome impairs the detection of disease-causing variants, in particular in transcripts only expressed under certain conditions. These transcripts are often lacking from reference transcript sets, such as Ensembl/GENCODE and RefSeq, and could be relevant for establishing genetic diagnoses. We present SUsPECT (Solving Unsolved Patient Exomes/gEnomes using Custom Transcriptomes), a pipeline based on the Ensembl Variant Effect Predictor (VEP) to predict variant impact on custom transcript sets, such as those generated by long-read RNA-sequencing, for downstream prioritization. Our pipeline predicts the functional consequence and likely deleteriousness scores for missense variants in the context of novel open reading frames predicted from any transcriptome. We demonstrate the utility of SUsPECT by uncovering potential mutational mechanisms of pathogenic variants in ClinVar that are predicted to be benign using the reference transcript annotation. In further support of SUsPECT's utility, we identified an enrichment of immune-related variants predicted to have a more severe molecular consequence when annotating with a newly generated transcriptome from stimulated immune cells instead of the reference transcriptome. Our pipeline outputs crucial information for further prioritization of potentially disease-causing variants for any disease and will become increasingly useful as more long-read RNA sequencing datasets become available.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513307v1?rss=1
Authors: Hao, B., Kovacs, I.
Abstract: As the current best practice, an experimental network dataset is validated by showing significant overlap with a gold standard network. Significance is assessed by comparison to a negative benchmark, often a randomized version of the same gold standard. While such analysis can reliably indicate the presence of signal, it is ill-suited to assess how much signal there is. As an alternative, here we introduce a positive statistical benchmark corresponding to the best-case scenario, capturing the maximum possible overlap between two networks. Such a positive benchmark can be efficiently generated in a maximum entropy framework and opens the way to assess if the observed overlap is significantly different from the best-case scenario. In combination with the negative benchmark, we provide a normalized overlap score Normlap. As an application, we compare molecular and functional networks, resulting in an agreement network of human as well as yeast network datasets. Although the number of shared interactions between most networks within the same organism is relatively small, we find that it is typically close to the best-case scenario. This paradox is resolved by the underlying degree inconsistency, meaning that highly connected hubs in one network often have small degrees in another, limiting the potential overlap. Furthermore, we illustrate how Normlap improves the quality assessment of experimental networks, fostering the creation of future high-quality networks.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.22.513369v1?rss=1
Authors: Qin, J., Liu, X., Liu, Y., Wei, P., Kangzhu, Y., Zhong, J., Wang, J.
Abstract: With the development of molecular biology and genetics, deep sequencing technology has become the main way to discover genetic variation and reveal the molecular structure of genome. Due to the complexity of the whole genome segment structure, a large number of missing genotypes have appeared after sequencing, and these missing genotypes can be imputed by genotype imputation method. With the in-depth study of genotype imputation methods, computational intensive and computationally efficient imputation software come into being. Beagle software, as an efficient imputation software, is widely used because of its advantages of low memory consumption, fast running speed and relatively high imputation accuracy. K-Means clustering can divide individuals with similar population structure into a class, so that individuals in the same class can share longer haplotype fragments. Therefore, combining K-Means clustering algorithm with Beagle software can improve the interpolation accuracy. The Beagle and KBeagle method was used to compare the imputation efficiency. The KBeagle method presents a higher imputation matching rate and a shorter computing time. In the genome selection and heritability estimated section, the genotype dataset after imputed, unimputed, and with real genotype show similar prediction accuracy. However the estimated heritability using genotype dataset after imputed is closer to the estimation by the dataset with real genotype. We generated a compounds and efficient imputation method, which presents valuable resource for improvement of imputation accuracy and computing time. We envisage the application of KBeagle will be focus on the livestock sequencing study under strong genetic structure.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513222v1?rss=1
Authors: Jardillier, R., Koca, D., Chatelain, F., Guyon, L.
Abstract: (1) Background: tumor profiling enables patient survival prediction. The two essential parameters to be calibrated when designing a study based on tumor profiles from a cohort are the sequencing depth of RNA-seq technology and the number of patients. This calibration is carried out under cost constraints, and a compromise has to be found. In the context of survival data, the goal of this work is to benchmark the impact of the number of patients and of the sequencing depth of miRNA-seq and mRNA-seq on the predictive capabilities for both the Cox model with elastic net penalty and random survival forest. (2) Results: we first show that the Cox model and random survival forest provide comparable prediction capabilities, with significant differences for some cancers. Second, we demonstrate that miRNA and/or mRNA data improve prediction over clinical data alone. mRNA-seq data leads to slightly better prediction than miRNA-seq, with the notable exception of lung adenocarcinoma for which the tumor miRNA profile shows higher predictive power. Third, we demonstrate that the sequencing depth of RNA-seq data can be reduced for most of the investigated cancers without degrading the prediction abilities, allowing the creation of independent validation sets at lower cost. Finally, we show that the number of patients in the training dataset can be reduced for the Cox model and random survival forest, allowing the use of different models on different patient subgroups. (3) Availability: R script is available at https://github.com/remyJardillier/Survival_seq_depth
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.24.513352v1?rss=1
Authors: Yu, Q., Van Minsel, P., Galle, E., Thienpont, B.
Abstract: Novel single cell RNA-seq analysis combined with CRISPR screens enables the high- throughput characterization of transcriptional changes caused by genetic perturbations. Dedicated software to annotate CRISPR guide RNA (gRNA) libraries and associate them with single cell transcriptomes are however lacking. Here, we generated a CRISPR droplet sequencing dataset. We demonstrate that the current default tool fails to detect mutant gRNAs. We therefore developed GiRAFR, a pysam-based software tool to characterize intact and mutant gRNAs. We show that mutant gRNAs are dysfunctional, and failure to detect and annotate them leads to an inflated estimate of the number of untransformed cells as well as an underestimated multiplet frequency. These findings are mirrored in publicly available datasets, where we find that up to 34 % of cells are transduced with a mutant gRNA. Applying GiRAFR hence stands to improve the annotation and quality of single cell CRISPR screens.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513318v1?rss=1
Authors: Moshiri, N.
Abstract: Motivation: In viral molecular epidemiology, reconstruction of consensus genomes from sequence data is critical for tracking mutations and variants of concern. However, storage of the raw sequence data can become prohibitively large, and computing consensus genome from sequence data can be slow and requires bioinformatics expertise. Results: ViReaDB is a user-friendly database system for compactly storing viral sequence data and rapidly computing consensus genome sequences. From a dataset of 1 million trimmed mapped SARS-CoV-2 reads, it is able to compute the base counts and the consensus genome in 16 minutes, store the reads alongside the base counts and consensus in 50 MB, and optionally store just the base counts and consensus (without the reads) in 300 KB. Availability: ViReaDB is freely available on PyPI (https://pypi.org/project/vireadb) and on GitHub (https://github.com/niemasd/ViReaDB) as an open-source Python software project.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513156v1?rss=1
Authors: Zhu, W., Zhu, W.
Abstract: Plant-specific BURP genes are involved in most aspects of plant development and in diverse stress responses. However, there has been no comprehensive and systematic research on the wheat (Triticum aestivum L.) BURP gene family. We comprehensively identified 54 BURP genes in wheat and 12, 23, and 32 BURP genes, respectively, in its three progenitor species, Triticum urartu, Aegilops tauschii, and Triticum dicoccoides. By phylogenetic analysis, we classified the wheat BURP genes into five subfamilies: BURP VI, BURP VII, RD22-like, PG1 beta-like, and the newly defined BURP VIII. BURP genes were distributed unevenly on 20 chromosomes, and 33 (61%) were anchored in the distal chromosome segments. Analysis of gene duplication events showed that segmental duplication was the main contributor to the expansion of this gene family in wheat. Assessment of tissue-specific and stress-induced expression indicated that most BURP members are heavily involved in plant development and responses to various stress conditions. RNA-seq data revealed ten TaBURP genes expressed specifically in spikes.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513259v1?rss=1
Authors: Domanskyi, S., Jocoy, E. L., Srivastava, A., Bult, C. J.
Abstract: Motivation: Drug synergy prediction is a complex problem typically approached with machine learning techniques using molecular data, pharmacological data, and knowledge of biological-interaction networks. The recently published Cancer Drug Atlas (CDA) uses a logistic regression model to predict a binary synergy outcome in cell-line models by utilizing drug target information, knowledge of genes mutated in each model, and the models monotherapy drug sensitivity. However, we observed low performance, 0.33, of the CDA measured by Pearson correlation of predicted versus measured sensitivity when we evaluated datasets from six studies that were not considered during the development of the CDA. Here we describe improvements to the CDA algorithm, the Augmented CDA, that improved performance by 71% and robustness to dataset variations in drug response values. Results: We augmented the drug-synergy prediction-modeling approach CDA described in Narayan et al. by applying a random forest regression and optimization via cross-validation hyper-parameter tuning. We benchmarked the performance of our Augmented CDA (ACDA) compared to the original CDA algorithm using datasets from DrugComb, an open-access drug-combination screening data resource. The ACDA's performance is 71% higher than that of the CDA when trained and validated on the same dataset spanning ten tissues. The ACDA performs marginally better (6% increase) than the CDA when trained on one dataset and validated on another dataset in 22 cases that cover seven tissues. We also compared the performance of ACDA to one of the winners of the DREAM Drug Combination Prediction Challenge (Mikhail Zaslavskiys algorithm which we denoted as EN). The performance of EN was smaller than that of the ACDA in 15 out of 19 cases. In addition to data from cell lines, we also trained the ACDA algorithm on Novartis Institutes for BioMedical Research PDX encyclopedia (NIBR PDXE) data and generated sensitivity predictions for the cases where drug-combination tumor-volume measurements were unavailable. Finally, we developed an approach to visualize synergy-prediction data using dendrograms and heatmaps instead of the Voronoi diagrams used in the CDA. The latter has a complex algorithmic realization and no publicly available implementation, whereas the ACDA visualization approach is more transparent and has open access. We implemented and wrapped the ACDA algorithm in an easy-to-use python package available from PyPI. Availability: The source code is available at https://github.com/TheJacksonLaboratory/drug-synergy, and the software package can be installed directly from PyPI using pip.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512904v1?rss=1
Authors: Yang, K. L., Yu, F., Teo, G. C., Demichev, V., Ralser, M., Nesvizhskii, A. I.
Abstract: Peptide identification in liquid chromatography-tandem mass spectrometry (LC-MS/MS) experiments relies on computational algorithms for matching acquired MS/MS spectra against sequences of candidate peptides using database search tools, such as MSFragger. Here, we present a new tool, MSBooster, for rescoring peptide-to-spectrum matches using additional features incorporating deep learning-based predictions of peptide properties, such as LC retention time, ion mobility, and MS/MS spectra. We demonstrate the utility of MSBooster, in tandem with MSFragger and Percolator, in several different workflows, including nonspecific searches (immunopeptidomics), direct identification of peptides from data independent acquisition data, single-cell proteomics, and data generated on an ion mobility separation-enabled timsTOF MS platform. MSBooster is fast, robust, and fully integrated into the widely used FragPipe computational platform.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.18.512676v1?rss=1
Authors: Jain, S., Chouzenoux, E., Kumar, K., Majumdar, A.
Abstract: Co-administration of two or more drugs simultaneously can result in adverse drug reactions. Identifying drug-drug interactions (DDIs) is necessary, especially for drug development and for repurposing old drugs. DDI prediction can be viewed as a matrix completion task, for which matrix factorization (MF) appears as a suitable solution. This paper presents a novel Graph Regularized Probabilistic Matrix Factorization (GRPMF) method, which incorporates expert knowledge through a novel graph-based regularization strategy within an MF framework. An efficient and sounded optimization algorithm is proposed to solve the resulting non-convex problem in an alternating fashion. The performance of the proposed method is evaluated through the DrugBank dataset, and comparisons are provided against state-of-the-art techniques. The results demonstrate the superior performance of GRPMF when compared to its counterparts.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512741v1?rss=1
Authors: Ding, J., Wen, H., Tang, W., Liu, R., Li, Z., Venegas, J., Su, R., Molho, D., Jin, W., Zuo, W., Wang, Y., Xie, Y., Tang, J.
Abstract: In the realm of single-cell analysis, computational approaches have brought an increasing number of fantastic prospects for innovation and invention. Meanwhile, it also presents enormous hurdles to reproducing the results of these models due to their diversity and complexity. In addition, the lack of gold-standard benchmark datasets, metrics, and implementations prevents systematic evaluations and fair comparisons of available methods. Thus, we introduce the DANCE platform, the first standard, generic, and extensible benchmark platform for accessing and evaluating computational methods across the spectrum of benchmark datasets for numerous single-cell analysis tasks. Currently, DANCE supports 3 modules and 8 popular tasks with 32 state-of-art methods on 21 benchmark datasets. People can easily reproduce the results of supported algorithms across major benchmark datasets via minimal efforts (e.g., only one command line). In addition, DANCE provides an ecosystem of deep learning architectures and tools for researchers to develop their own models conveniently. The goal of DANCE is to accelerate the development of deep learning models with complete validation and facilitate the overall advancement of single-cell analysis research. DANCE is an open-source python package that welcomes all kinds of contributions. All resources are integrated and available at https://omicsml.ai/.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.17.512224v1?rss=1
Authors: Zulfiqar, M., Gadelha, L., Steinbeck, C., Sorokina, M., Peters, K.
Abstract: Mapping the chemical space of compounds to chemical structures remains a challenge in metabolomics. Despite the advancements in untargeted Liquid Chromatography Mass Spectrometry (LCMS) to achieve a high-throughput profile of metabolites from complex biological resources, only a small fraction of these metabolites can be annotated with confidence. Many novel computational methods and tools have been developed to enable chemical structure annotation to known and unknown compounds such as in-silico generated spectra and molecular networking. We have developed an automated and reproducible Metabolome Annotation Workflow (MAW) for untargeted metabolomics data implemented in R and Python to assist the metabolic annotation further by combining Tandem Mass Spectrometry (MS2) input data pre-processing and different computational approaches to give a reliable structure candidate. MAW takes the LC-MS2 spectra as input and generates a list of putative candidates from spectral and compound libraries. The libraries are integrated via an R package called Spectra and through the CLI (Command-Line-Interface) version of the metabolite annotation tool SIRIUS as part of the R segment of the workflow (MAW-R). The final candidate selection is performed using the cheminformatics tool RDKit in the Python segment (MAW-Py). Furthermore, each feature is assigned a chemical structure and can be imported to a chemical structure similarity network. Following FAIR (Findable, Accessible, Interoperable, Reusable) principles, MAW is available as Docker images. The source code and documentation are available on the GitHub repository zmahnoor14/MAW and the version of the workflow presented in this article has a DOI (10.5281/zenodo.7148450) associated with it.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.17.512527v1?rss=1
Authors: Wang, J., Wan, Y.-W., Al-Ouran, R., Liu, Z.
Abstract: Scientists have learned much about gene expression in the past few years. Integrating large-scale gene-expression data will significantly bolster researchers' understanding of biology and diseases. While we currently use co-expression models to discover gene-gene associations in RNA seq data, integrating RNA seq data generated from different experiments can have a "batch effect", which decreases data quality and makes it harder to glean definitive gene relationships from co-expression models. The biological relationship can also be context-dependent and non-linear, which make them difficult to detect using current co-expression approaches. Here, we propose a co-perturbation model that better identifies gene-gene associations in integrated data and reveals non-linear correlation between genes.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.17.512602v1?rss=1
Authors: Mao, H., Jia, M., Di, M., Zhang, K., Benos, P. V.
Abstract: As the available sequencing data modalities increase, so does the potential biological insight that they are able to provide. Most existing methods to integrate co-profiled single-cell multi-omics data focus only on learning representations that capture stationary and shared information among these modalities. Current methods do not account for time-dependent and modality-specific information delineating cell states and subtypes, nor do they consider dynamics resulting from causal relations among modalities. For example, open chromatin may cause active transcription; however, it is also possible that gene expression responses lag behind changes in chromatin accessibility. To account for this time lag, the epigenome and transcriptome relationship can be characterized as "coupled" (changing dependently) or "decoupled" (changing independently). We propose the framework HALO (Hierarchical cAusal representation Learning for Omics data), which adopts a causal approach to model these non-stationary causal relations using independent changing mechanisms in co-profiled single-cell ATAC- and RNA-seq data. Our model factorizes these two modalities into both coupled and decoupled latent representations, allowing us to identify the dynamic interplay between chromatin accessibility and transcription through temporal modulations. In blood lineage and developing mouse brain data, where the balance between proliferation and differentiation is tightly regulated, HALO distinguishes between coupled and decoupled genes and links them with disparate processes that constitute these two complementary states.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.20.513077v1?rss=1
Authors: Asif, M., Martiniano, H., Lamurias, A., Kausar, S., Couto, F.
Abstract: Complex diseases such as neurodevelopmental disorders (NDDs) lack biological markers for their diagnosis and are phenotypically heterogeneous, which makes them difficult to diagnose at early-age. The genetic heterogeneity corresponds to their clinical phenotype variability and, because of this, complex diseases exhibit multiple etiologies. The multi-etiological aspects of complex-diseases emerge from distinct but functionally similar group of genes. Different diseases sharing genes of such groups show related clinical outcomes that further restrict our understanding of disease mechanisms, thus, limiting the applications of personalized medicine or systems biomedicine approaches to complex genetic disorders. Here, we present an interactive and user-friendly application, DGH-GO that allows biologists to dissect the genetic heterogeneity of complex diseases by stratifying the putative disease-causing genes into clusters that may lead to or contribute to a specific disease traits development. The application can also be used to study the shared etiology of complex-diseases. DGH-GO creates a semantic similarity matrix of putative disease-causing genes or known-disease genes for multiple disorders using Gene Ontology (GO). The resultant matrix can be visualized in a 2D space using different dimension reduction methods (T-SNE, Principal component analysis and Principal coordinate analysis). Functional similarities assessed through GO and semantic similarity measure can be used to identify clusters of functionally similar genes that may generate a disease specific traits. This can be achieved by employing four different clustering methods (K-means, Hierarchical, Fuzzy and PAM). The user may change the clustering parameters and see their effect on stratification results immediately. DGH-GO was applied to genes disrupted by rare genetic variants in Autism Spectrum Disorder (ASD) patients. The analysis confirmed the multi-etiological nature of ASD by identifying the four clusters that were enriched for distinct biological mechanisms and phenotypic terms. In the second case study, the analysis of genes shared by different NDDs showed that genes involving in multiple disorders tend to aggregate in similar clusters, indicating a possible shared etiology. In summary, functional similarities, dimension reduction and clustering methods, coupled with interactive visualization and control over analysis allows biologists to explore and analyze their datasets without requiring expert knowledge on these methods. The source code of proposed application is available at https://github.com/Muh-Asif/DGH-GO Keywords: Unsupervised learning, dimension reduction, gene ontology, functionally similarities, semantic similarity, genetic heterogeneity, neurodevelopment disorders
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512926v1?rss=1
Authors: Ghaddar, B., De, S.
Abstract: We developed Census, an automated, hierarchical cell-type identification method for scRNA-seq data that can deeply annotate normal cells in mammalian tissues and identify malignant cells and their likely cell of origin. When benchmarked on 44 atlas-scale normal and cancer, human and mouse tissues, Census significantly outperforms state-of-the-art methods across multiple metrics. Census is a fast and fully automated method, although users can seamlessly train their own models for customized applications.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512881v1?rss=1
Authors: Hoffmann, M., Trummer, N., Jankowski, J., Lee, H. K., Willruth, L.-L., Yuan, K., Lazareva, O., Baumgarten, N., Schmidt, F., Baumbach, J., Schulz, M. H., Blumenthal, D. B., Hennighausen, L., List, M.
Abstract: Background Eukaryotic gene expression is controlled by cis-regulatory elements (CREs) including promoters and enhancers which are bound by transcription factors (TFs). Differential expression of TFs and their putative binding sites on CREs cause tissue and developmental-specific transcriptional activity. Consolidating genomic data sets can offer further insights into the accessibility of CREs, TF activity, and thus gene regulation. However, the integration and analysis of multi-modal data sets are hampered by considerable technical challenges. While methods for highlighting differential TF activity from combined ChIP-seq and RNA-seq data exist, they do not offer good usability, have limited support for large-scale data processing, and provide only minimal functionality for visual result interpretation. Results We developed TF-Prioritizer, an automated java pipeline to prioritize condition-specific TFs derived from multi-modal data. TF-Prioritizer creates an interactive, feature-rich, and user-friendly web report of its results. To showcase the potential of TF-Prioritizer, we identified known active TFs (e.g., Stat5, Elf5, Nfib, Esr1), their target genes (e.g., milk proteins and cell-cycle genes), and newly classified lactating mammary gland TFs (e.g., Creb1, Arnt). Conclusion TF-Prioritizer accepts ChIP-seq and RNA-seq data, as input and suggests TFs with differential activity, thus offering an understanding of genome-wide gene regulation, potential pathogenesis, and therapeutic targets in biomedical research.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.18.512682v1?rss=1
Authors: Pellow, D., Pu, L., Ekim, B., Kotlar, L., Berger, B., Shamir, R., Orenstein, Y.
Abstract: Minimizers are ubiquitously used in data structures and algorithms for efficient searching, mapping, and indexing of high-throughput DNA sequencing data. Minimizer schemes select a minimum k-mer in every L-long sub-sequence of the target sequence, where minimality is with respect to a predefined k-mer order. Commonly used minimizer orders select more k-mers overall than necessary and therefore provide limited improvement to runtime and memory usage of downstream analysis tasks. The recently introduced universal k-mer hitting sets produce minimizer orders resulting in fewer selected k-mers. Unfortunately, generating compact universal k-mer hitting sets is currently infeasible for k greater than 13, and thus cannot help in the many applications that need minimizers of larger k. Here, we close this gap by introducing decycling set-based minimizer orders. We define new orders based on minimum decycling sets, which are guaranteed to hit any infinitely long sequence. We show that in practice these new minimizer orders select a number of k-mers comparable to that of minimizer orders based on universal k-mer hitting sets, and can also scale up to larger k. Furthermore, we developed a query method that avoids the need to keep the k-mers of a decycling set in memory, which enables the use of these minimizer orders for any value of k. We expect the new decycling set-based minimizer orders to improve the runtime and memory usage of algorithms and data structures in high-throughput DNA sequencing analysis.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513099v1?rss=1
Authors: McWhite, C. D., Singh, M.
Abstract: Multiple sequence alignment is a critical step in the study of protein sequence and function. Typically, multiple sequence alignment algorithms progressively align pairs of sequences and combine these alignments with the aid of a guide tree. These alignment algorithms use scoring systems based on substitution matrices to measure amino-acid similarities. While successful, standard methods struggle on sets of proteins with low sequence identity - the so-called twilight zone of protein alignment. For these difficult cases, another source of information is needed. Protein language models are a powerful new approach that leverage massive sequence datasets to produce high-dimensional contextual embeddings for each amino acid in a sequence. These embeddings have been shown to reflect physicochemical and higher-order structural and functional attributes of amino acids within proteins. Here, we present a novel approach to multiple sequence alignment, based on clustering and ordering amino acid contextual embeddings. Our method for aligning semantically consistent groups of proteins circumvents the need for many standard components of multiple sequence alignment algorithms, avoiding initial guide tree construction, intermediate pairwise alignments, gap penalties, and substitution matrices. The added information from contextual embeddings leads to higher accuracy alignments for structurally similar proteins with low amino-acid similarity. We anticipate that protein language models will become a fundamental component of the next generation of algorithms for generating MSAs.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.20.510630v1?rss=1
Authors: Ahmadian, M., Rickert, C., Minic, A., Wrobel, J., Bitler, B. G., Zing, F., Angelo, M., Hsieh, E. W., Ghosh, D., Jordan, K. R.
Abstract: Multiplex imaging is a powerful tool to analyze the structural and functional states of cells in their morphological and pathological contexts. However, hypothesis testing with multiplex imaging data is a challenging task due to the extent and complexity of the information obtained. Various computational pipelines have been developed and validated to extract knowledge from specific imaging platforms. A common problem with customized pipelines is their reduced applicability across different imaging platforms: Every multiplex imaging technique exhibits platform-specific characteristics in terms of signal-to-noise ratio and acquisition artifacts that need to be accounted for to yield reliable and reproducible results. We propose a pixel classifier-based image preprocessing step that aims to minimize platform-dependency for all multiplex image analysis pipelines. Signal detection and noise reduction as well as artifact removal can be posed as a pixel classification problem in which all pixels in multiplex images can be assigned to two general classes of either I) signal of interest or II) artifacts and noise. The resulting feature representation maps contain pixel-accurate representations of the input data, but exhibit significantly increased signal-to-noise ratios with normalized pixel values as output data. We demonstrate the validity of our proposed image preprocessing approach by comparing the results of two well-accepted and widely-used image analysis pipelines.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.20.513070v1?rss=1
Authors: Niccolai, N.
Abstract: It has been recently suggested that amino acid replacements with Gly can modify the shape of protein surfaces and, hence, protein dynamics and functions. We have browsed ClinVar, the database of all the reported variants of clinical relevance, to identify all the proteins having missense X/Gly mutations that determine Mendelian disorders. We have found 959 benign and 875 pathogenic X/Gly substitutions. Pathogenicity origins were initially searched in the distribution profiles of replaced amino acids. These profiles indicate that Mendelian disorders including Gly-replacements arise mainly from substitutions of amino acids bearing bulky hydrophobic side chains, thus reducing protein core stability. In the case mutated proteins were structurally defined, we could give a deeper insight into pathogenicity mechanisms, checking whether Gly-mutations altered protein shapes, modifying water surface dynamics and, hence, the physiological protein-protein interaction processes. In several cases, indeed, we have found that pathological Gly-mutants present additional surface pockets, suggesting that the new pockets could be the target of a pharmacological strategy for Mendelian disorder remediation.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.18.512665v1?rss=1
Authors: Narendra, T., Visona, G., Cardona, C. d. J., Schweikert, G.
Abstract: Epigenetic mechanisms coordinate packaging, accessibility and read-out of the DNA sequence within the chromatin context. They significantly contribute to the regulation of gene expression. Thus, they play fundamental roles during differentiation on the one hand and maintenance and propagation of cell identity on the other. Epigenetic malfunctioning is associated with a large range of diseases, from neurodevelopmental disorders to cancer progression. In humans, hundreds of known epigenetic factors and complexes are involved in establishing covalent modifications on the DNA sequence itself and on associated histone proteins. Within the cellular context, the resulting combinatorial epigenomic patterns are neither established nor interpreted independently of each other and therefore exhibit high correlations in a region-specific manner. Post-translational modifications of histone proteins can be analysed using Chromatin Immunoprecipitation followed by sequencing (ChIP-Seq). Often, several assays for a number of different histone modifications are performed as part of the same experimental design. These measurements are, however, confounded by shared biases including chromatin accessibility, PCR amplification and mappability. Existing computational methods analyse each histone modification separately, while often also merging biological or technical replicates. We introduce DecoDen, a new approach that leverages replicates and multi-histone ChIP-Seq experiments for a fixed cell type to learn and remove shared biases. DecoDen (Deconvolve and Denoise) consists of two major steps: We use non-negative matrix factorisation (NMF) to learn a joint cell-type specific signal. Half-sibling regression (HSR) is then used to correct for the cell-type specific biases in the histone modification signals. We demonstrate that DecoDen is a robust and interpretable method that enables the unbiased discovery of subtle peaks, which are particularly important in an individual-specific context.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512965v1?rss=1
Authors: Devi, N. L., Sharma, N., Raghava, G. P. S.
Abstract: Interleukin-5 (IL-5) is the key cytokine produced by T-helper, eosinophils, mast and basophils cells. It can act as an enticing therapeutic target due to its pivotal role in several eosinophil-mediated diseases. Though numerous methods have been developed to predict HLA binders and cytokines-inducing peptides, no method was developed for predicting IL-5 inducing peptides. All models in this study have been trained, tested and validated on experimentally validated 1907 IL-5 inducing and 7759 non-IL-5 inducing peptides obtained from IEDB. First, alignment-based methods have been developed using similarity and motif search. These alignment-based methods provide high precision but poor coverage. In order to overcome this limitation, we developed machine learning-based models for predicting IL-5 inducing peptides using a wide range of peptide features. Our random-forest model developed using selected 250 dipeptides achieved the highest performance among alignment-free methods with AUC 0.75 and MCC 0.29 on validation dataset. In order to improve the performance, we developed an ensemble or hybrid method that combined alignment-based and alignment-free methods. Our hybrid method achieved AUC 0.94 with MCC 0.60 on validation/ independent dataset. The best model developed in this study has been incorporated in the web server IL5pred (https://webs.iiitd.edu.in/raghava/il5pred/).
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513201v1?rss=1
Authors: Lee, H., Welch, J. D.
Abstract: Gene expression and morphology both play a key role in determining the types and functions of cells, but the relationship between molecular and morphological features is largely uncharacterized. We present MorphNet, a computational approach that can draw pictures of a cell's morphology from its gene expression profile. Our approach leverages paired morphology and molecular data to train a neural network that can predict nuclear or whole-cell morphology from gene expression. We employ state-of-the-art data augmentation techniques that allow training using as few as 10^3 images. We find that MorphNet can generate novel, realistic morphological images that retain the complex relationship between gene expression and cell appearance. We then train MorphNet to generate nuclear morphology from gene expression using brain-wide MERFISH data. In addition, we show that MorphNet can generate neuron morphologies with realistic axonal and dendritic structures. MorphNet generalizes to unseen brain regions, allowing prediction of neuron morphologies across the entire mouse isocortex and even non-cortical regions. We show that MorphNet performs meaningful latent space interpolation, allowing prediction of the effects of gene expression variation on morphology. Finally, we provide a web server that allows users to predict neuron morphologies for their own scRNA-seq data. MorphNet represents a powerful new approach for linking gene expression and morphology.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512902v1?rss=1
Authors: Danaher, P., Zhao, E., Yang, Z., Ross, D., Gregory, M., Reitz, Z., Kim, T. K., Baxter, S., Jackson, S., He, S., Henderson, D. A., Beechem, J. M.
Abstract: Accurate cell typing is fundamental to analysis of spatial single-cell transcriptomics, but legacy scRNA-seq algorithms can underperform in this new type of data. We have developed a cell typing algorithm, Insitutype, designed for statistical and computational efficiency in spatial transcriptomics data. Insitutype is based on a likelihood model that weighs the evidence from every expression value, extracting all the information available in each cell's expression profile. This likelihood model underlies a Bayes classifier for supervised cell typing, and an Expectation-Maximization algorithm for unsupervised and semi-supervised clustering. Insitutype also leverages alternative data types collected in spatial studies, such as cell images and spatial context, by using them to inform prior probabilities of cell type calls. We demonstrate rapid clustering of millions of cells and accurate fine-grained cell typing of kidney and non-small cell lung cancer samples.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513161v1?rss=1
Authors: Kaynar, G., Cakmakci, D., Bund, C., Todeschi, J., Namer, I. J., Cicek, A. E.
Abstract: Online assessment of the tumor pathology during surgery is an important task to give feedback to the surgeon who can be more liberal or conservative in the resection based on the input. While there are methods that perform metabolomics-based online tumor grade prediction, their performance and model complexities are limited by the small dataset sizes. Here, we propose a pathway-informed deep learning model, PiDeel, to perform survival analysis simultaneously for a better prognostic assessment. We show that incorporating the pathway information into the model architecture and using the multitask learning framework reduce the model complexity and enable us to use deeper architectures with better performance. We show that PiDeel improves the grade prediction performance of the state-of-the-art in terms of the Area Under the ROC Curve (AUC-ROC) by 3.38% and the Area Under the Precision-Recall Curve (AUC-PR) by 4.06% and survival analysis performance based on the time-dependent concordance index (c-index) by 2.06%. Analyzing the importance of input metabolites and neurons representing pathways provides insights into tumor metabolism. We foresee that using this model in the surgery room will help surgeons adjust the surgery plan on the fly and will result in improved prognosis due to optimized surgical procedures. The code is released at https://github.com/ciceklab/PiDeeL. The data used in this study is released at https://zenodo.org/record/7228791.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513175v1?rss=1
Authors: Li, H.
Abstract: Machine learning and deep learning are novel and trending approaches to solving real-world scientific problems. Graph machine learning is dedicated to performing learning methods, such as graph neural networks, on non-Euclidean data such as graphs. Molecules, with their natural graph structures, could be analyzed by such method. In this work, we carry out the performance evaluation regarding to learning results as well as time consumed, speedup, and efficiency using different types of neural network structures and distributed training pipeline implementations. Besides, the reasons lead to an unideal performance enhancement is investigated. Code availability at https://github.com/ htlee6/perf-analysis-dist-training-gnn.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513209v1?rss=1
Authors: Shi, Y., Yang, W., Lin, H., Han, L., Cai, A. J., Saraf, R., Lei, Y., Zhang, C.
Abstract: Cell-type biomarkers are useful in stem cell manufacture to monitor cell purification, cell quantity, and quality. However, the study on cell-type markers, specifically for stem cell manufacture, is limited. The emerging questions are which RNA transcripts can serve as biomarkers during stem cell culture, and what method can efficiently and accurately discover these biomarkers. We developed a scoring function system to identify RNA biomarkers with RNA-seq data. We applied the method to two data sets, one for extracellular RNAs (ex-RNAs) and the other for intracellular microRNAs (miRNAs). The data set have RNA-seq data of ex-RNAs from cell culture media for six different types of cells, including human embryonic stem cells. To get the RNA-seq data from intracellular miRNAs, we cultured three types of cells: human embryonic stem cells (H9), neural stem cells (NSC), hESC-derived endothelial cells (EC) and conducted small RNA-seq to their intracellular miRNAs. Using these data, we identified a set of ex-RNAs/smRNAs as candidates of biomarkers for different types of cells for cell manufacture. We also used deep-learning based prediction methods and simulated data to validate these discovered biomarkers.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513181v1?rss=1
Authors: Luo, R., Lin, Y.
Abstract: With the high mutation rate in viruses, a mixture of closely related viral strains (called viral quasispecies) often co-infect an individual host. Reconstructing indi- vidual strains from viral quasispecies is a key step to characterizing the viral population, revealing strain-level genetic variability, and providing insights into biomedical and clin- ical studies. Reference-based approaches of reconstructing viral strains suffer from the lack of high-quality references due to high mutation rates and biased variant calling introduced by a selected reference. De novo methods require no references but face chal- lenges due to errors in reads, the high similarity of quasispecies, and uneven abundance of strains. In this paper, we propose VStrains, a de novo approach for reconstructing strains from viral quasispecies. VStrains incorporates contigs, paired-end reads, and coverage infor- mation to iteratively extract the strain-specific paths from assembly graphs. We bench- mark VStrains against multiple state-of-the-art de novo and reference-based approaches on both simulated and real datasets. Experimental results demonstrate that VStrains achieves the best overall performance on both simulated and real datasets under a com- prehensive set of metrics such as genome fraction, duplication ratio, NGA50, error rate, etc. VStrains is publicly available at https://github.com/MetaGenTools/VStrains.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.20.512704v1?rss=1
Authors: Ghimire, S., Sahukhal, S., Shrestha, A., Adhikari, S., Subedi, S., Budha, K. R., Aryal, P.
Abstract: Mutation, reassortment and recombination have led to the evolution and the emergence of more pathogenic and new subtypes of influenza virus. The surge of highly mutated viruses has prompted the need of coherent solution for the so called ''medical holocaust'' viral outbreaks.The genotype 4 of EAH1N1 strain has been circulating in the swine population as a dominant genotype, exhibiting even human to human transmission. This has risen the possibility of causing another global health threat as a lethal viral outbreak in the future. The Computer Aided Drug Discovery (CADD) could be a prudent mechanism to develop new drug candidates against such disease for its mitigation. In this regard, the computational in silico methods had been envisaged in this research for the prediction of lead compounds against the selected proteins of EA H1N1 G4 strain, namely Haemagglutinin (HA) and Polymerase acidic protein(PA). The research focused on the selection of the target viral protein and molecular docking for the identification of putative ligands. It was followed by the identification of the probable mutations and assessment of effectiveness of identified drugs against their respective targets. Total of 3 compounds Enalapril, Enalaprilat and Ivabradine have been identified as a potential inhibitor of HA and PA protein that were prioritized on the basis of preference index parameter and binding energy of compound with the respective target. Besides, the probable mutations in each target protein in future were predicted and all these 3 top hits were found to be effective against mutated variant of these proteins. Thus, Enalapril, Enalaprilat and Ivabradine could be the lead compounds to explore further as multi target inhibiting drugs against wild and mutant variant of target proteins. Keywords:CADD, Haemagglutinin, Molecular docking ,Pandemics ,Polymerase Acidic protein
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.20.513017v1?rss=1
Authors: Cornet, L., Durieu, B., Baert, F., D'hooge, E., Colignon, D., Meunier, L., Lupo, V., Cleenwerck, I., Daniel, H.-M., Rigouts, L., Sirjacobs, D., Declerck, S., Vandamme, P., Wilmotte, A., Baurain, D., Becker, P.
Abstract: Background: Microbial culture collections play a key role in taxonomy by studying the diversity of their accessions and providing well characterized strains to the scientific community for fundamental and applied research. These microbial resource centers thus need to implement new standards in species delineation, including whole-genome sequencing and phylogenomics. In this context, the genomic needs of the Belgian Coordinated Collections of Microorganisms (BCCM) were studied, resulting in the GEN-ERA toolbox. The latter is a unified cluster of bioinformatic workflows dedicated to both bacteria and small eukaryotes (i.e. yeasts). Findings: This public toolbox allows researchers without a specific training in bioinformatics to perform robust phylogenetic analyses. Hence, it facilitates all steps from genome downloading and quality assessment, including genomic contamination estimation, to tree reconstruction. It also offers workflows for average nucleotide identity comparisons and metabolic modeling. Technical details: Nextflow workflows are launched by a single command and are available on the GEN-ERA GitHub repository (https://github.com/Lcornet/GENERA). All the workflows are based on Singularity containers to increase reproducibility. Testing: The toolbox was developed for a diversity of microorganisms, including bacteria and fungi. It was further tested on an empirical dataset of 18 (meta)genomes of early-branching Cyanobacteria, providing the most up-to-date phylogenomic analysis of the Gloeobacterales order, the first group to diverge in the evolutionary tree of Cyanobacteria. Conclusion: The GEN-ERA toolbox can be used to infer completely reproducible comparative genomic and metabolic analyses on prokaryotes and small eukaryotes. Although designed for routine bioinformatics of culture collections, it can also be useful for other applications, as shown by our case study on Gloeobacterales.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.492777v1?rss=1
Authors: Palamin, M., Frisinghelli, A., Offer, E., Zampieri, G., Basile, A., Campanaro, S.
Abstract: Motivation: Microorganisms have a pivotal role in ecology and human health and form complex networks where different species can interact and exchange a range of different compounds. Flux balance analysis can offer an insight into the production and the absorption of these metabolic compounds, but frequently results are difficult to visualise and interpret. Additionally, a clear understanding of the roles of microbial species in the community requires the integration of different information sources, including relative abundance, taxonomy and compounds exchange rate. Results: To fill-in this gap the command-line tool NEMETEX (NEtwork for METabolic EXchanges) was developed to provide a graphical representation of the metabolites exchanged, joined with interactive visualisation of numerical data. This approach can undoubtedly represent an easy way to investigate high-throughput results obtained from metagenomics and flux balance analysis, providing a more direct interpretation of the data. Availability and implementation: This program, accessory utilities, and their documentation are freely available at https://github.com/palakela/NEMETEX
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512911v1?rss=1
Authors: Friedlos, P., Gasser, L., Harris, E.
Abstract: Comparing measured and predicted chromatographic retention time can improve molecular structure assignment in applications such as coupled liquid chromatography-tandem mass spectrometry. We assess a range of different machine learning methods to predict hydrophobicity, a molecular property that can be used as a proxy for retention time. The performance of the models is evaluated on the benchmark Martel and SAMPL7 datasets. We find that more powerful models perform better when predicting in-sample but not necessarily when generalizing to out-of-sample molecular families. We also find that ensemble methods can outperform individual models. Additionally, a multitask learning model shows promise for improving the generalization ability of graph neural networks for hydrophobicity prediction. Finally, we discuss how the ability of graph neural networks to generalize for molecular property prediction could be improved further.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512838v1?rss=1
Authors: He, G., Chen, M., Bian, Y., Yang, E.
Abstract: Mapping biological information from peripheral 'surrogate' samples, especially transcriptomic information from blood, to tissue expression profiles has become an effective emerging alternative when invasive procedures are not ideal. However, existing approaches ignore hidden nonlinear expression relationships and disrupt tissue-shared intrinsic relevance, inevitably limiting predictive performance. Here, we propose a unified deep-learning-based multi-task learning framework, Multi-tissue Transcriptome Mapping (MTM), that enables the prediction of individualized expression profiles from any available tissue from an individual. By jointly leveraging individualized cross-tissue information through nonlinear neural networks, MTM achieves superior performance at both the sample level and the gene level with a large proportion of predictable genes. With the high accuracy of predicting the expression profiles of uncollected tissues and the ability to preserve individualized biological variations, MTM could facilitate the discovery of novel mechanisms and clinical biomarkers in biomedical research.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.20.513029v1?rss=1
Authors: Jiang, Y., Huo, M., Li, S.
Abstract: The adaptive immune response to foreign antigens is initiated by T-cell receptor (TCR) recognition on the antigens. Recent experimental advances have enabled the generation of a large amount of TCR data and their cognate antigenic targets, allowing machine learning models to predict the binding specificity of TCRs. In this work, we present TEINet, a deep learning framework that utilizes transfer learning to address this prediction problem. TEINet employs two separately trained encoders to transform TCR and epitope sequences into numerical vectors, which are subsequently fed into a fully connected neural network to predict their binding specificities. A major challenge for binding specificity prediction is the lack of a unified approach to sample negative data. Here, we first assess the current negative sampling approaches comprehensively and suggest that the Unified Epitope is the most suitable one. Subsequently, we compare TEINet with three baseline methods and observe that TEINet achieves an AUROC of 0.760, which outperforms baseline methods by 6.4-26%. Furthermore, we investigate the impacts of the pretraining step and notice that excessive pretraining can adversely affect model performance. Our results and analysis show that TEINet can make an accurate prediction using only the TCR sequence (CDR3$beta$) and the epitope sequence, providing novel insights to understand the interactions between TCRs and epitopes. TEINet is available at https://github.com/jiangdada1221/TEINet.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.19.512935v1?rss=1
Authors: Liu, T., Wang, Z.
Abstract: Chromatin interaction analysis by paired-end tag sequencing (ChIA-PET) can capture genome-wide chromatin interactions mediated by a specific DNA-associated protein. The ChIA-PET experiments have been applied to explore the key roles of different protein factors in chromatin folding and transcription regulation. However, compared with widely available Hi-C and ChIP-seq data, there are not many ChIA-PET datasets available in the literature. A computational method for accurately predicting ChIA-PET interactions from Hi-C and ChIP-seq data is needed that can save the efforts of performing wet-lab experiments. Here we present DeepChIA-PET, a supervised deep learning approach that can accurately predict ChIA-PET interactions by learning the latent relationships between ChIA-PET and two widely used data types: Hi-C and ChIP-seq. We trained our deep models with CTCF-mediated ChIA-PET of GM12878 as ground truth, and the deep network contains 40 dilated residual convolutional blocks. We first showed that DeepChIA-PET with only Hi-C as input significantly outperforms Peakachu, another computational method for predicting ChIA-PET from Hi-C but using random forests. We next proved that adding ChIP-seq as one extra input does improve the classification performance of DeepChIA-PET, but Hi-C plays a more prominent role in DeepChIA-PET than ChIP-seq. Our evaluation results indicate that our learned models can accurately predict not only CTCF-mediated ChIA-ET in GM12878 and HeLa but also non-CTCF ChIA-PET interactions, including RNA polymerase II (RNAPII) ChIA-PET of GM12878, RAD21 ChIA-PET of GM12878, and RAD21 ChIA-PET of K562. In total, DeepChIA-PET is an accurate tool for predicting the ChIA-PET interactions mediated by various chromatin-associated proteins from different cell types. DeepChIA-PET is publicly available at http://dna.cs.miami.edu/DeepChIA-PET/.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2022.10.21.513218v1?rss=1
Authors: Zhao, J., Pierre-both, J., Rodriguez-R, L. M., Konstantinidis, K. T.
Abstract: Genome search and/or classification is a key step in microbiome studies and has become more challenging due to the increasing number of available (reference) genomes in recent years and the fact that traditional methods do not scale well with larger databases. By combining a kmer hashing-based genomic distance metric (Probminhash) with a graph based nearest neighbor search (NNS) algorithm (called Hierarchical Navigable Small World Graphs), we developed a new program, GSearch, that is at least ten times faster than alternative tools for the same purposes while maintaining high accuracy. GSearch can identify/classify eight thousand query genomes against all available microbial and viral genomic species within several minutes on a personal laptop, using only ~6GB of memory. Further, GSearch can scale well with millions of database genomes based on a database splitting strategy. Therefore, GSearch solves a major bottleneck in current and future microbiome studies that require genome search and/or classification.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC