Sex differences in prevalence, severity, and genetic susceptibility exist for most common diseases. Most clinical outcomes and genetic data analyses are designed in sex-combined framework considering sex as a covariate, and only few studies analyze males and females separately, which fail to identify gene-by-sex interaction. Here, we propose a novel unified interpretable DL-based framework (SPIN) for sexual dimorphism analysis. SPIN identifies sex-specific and -shared risk loci that are often missed in sex-combined/-separate analysis. We also demonstrate that SPIN is interpretable for explaining how biological pathways contribute to sexual dimorphism and improve risk prediction, contributing to precision medicine initiatives.
- Source code: https://github.com/datax-lab/SPIN
Published in Briefings in Bioinformatics 2024, https://doi.org/10.1093/bib/bbae239