MTM: a multi-task learning framework to predict individualized tissue gene expression profiles

Abstract

Motivation: Transcriptional profiles of diverse tissues provide significant insights in both fundamental and translational researches, while transcriptome information is not always available for tissues that require invasive biopsies. Alternatively, predicting tissue expression profiles from more accessible “surrogate” samples, especially blood transcriptome, has become a promising strategy when invasive procedures are not practical. However, existing approaches ignore tissue-shared intrinsic relevance, inevitably limiting predictive performance. Results: We propose a unified deep learning-based multi-task learning framework, multi-tissue transcriptome mapping (MTM), enabling the prediction of individualized expression profiles from any available tissue of an individual. By jointly leveraging individualized cross-tissue information from reference samples through multi-task learning, MTM achieves superior sample-level and gene-level performance on unseen individuals. With the high prediction accuracy and the ability to preserve individualized biological variations, MTM could facilitate both fundamental and clinical biomedical research. Availability and implementation: MTM’s code and documentation are available upon publication on GitHub (https://github.com/yangence/ MTM).