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