Deep learning approach reveals multi-tissue connections in gene expression

In recent times, numerous studies have linked specific genes to complex diseases like cancer and diabetes, offering hope for significant advancements in medical treatments and drug discovery. However, these gene expression datasets are often disconnected and fail to reveal the intricate connections between different tissues, such as those between the brain and gut.

To address this, researchers from Vanderbilt University Medical Center and the University of Cambridge have developed a method called hypergraph factorization (HYFA) that predicts gene expression in hard-to-access tissues, like the brain, using more accessible tissues, such as whole blood. This deep-learning/machine-learning approach has unveiled shared gene expression patterns between the brain and gastrointestinal tract and validated certain blood genes as markers for Alzheimer’s disease in the brain.

By reconstructing and predicting gene expression across various tissues and cell types, this approach has the potential to expand our understanding of the molecular origins of complex traits. It enables biomarker discovery and drug repurposing research, as it harmonizes large molecular datasets generated by different researchers.

The study has also discovered and validated new genetic variants that can regulate gene levels in specific tissues and cell types, potentially underlying complex diseases and their comorbidities.

Eric Gamazon, an assistant professor of Medicine at VUMC and co-senior author with Pietro Liò of the University of Cambridge, is a leading expert in gene-expression data and a contributor to the Genotype-Tissue Expression (GTEx) project of the National Institutes of Health (NIH) Common Fund. Previous work by Gamazon and his colleagues has identified neuroendocrine and gastrointestinal contributors to psychiatric disorders.

The current study, co-authored by bioinformatics scientist Phillip Lin, takes the research to a systemwide view of human physiology, accelerating the integration of large-scale tissue and cell-type gene expression biorepositories as more resources are generated by studies, institutions, and consortia. The research holds the promise of significant advancements in medical understanding and potential treatments for various complex diseases.

Source: Vanderbilt University

Leave a Comment