Back

Genetic Profiling and Early Detection of Type 2 Diabetes Subtypes through Sex-Stratified GWAS and Explainable AI

Alonso Parrilla, L.; Perez-Elena, M. A.; Salem, M. Y.; Mashhadikhan, M.; Gaitan, N.; Satari, L.; Martin, R.; Piron, A.; Farre, X.; Blay, N.; Ruiz, L.; Lymperidou, A.; Salvoro, C.; de Cid, R.; Berral, J. L.; Gonzalez, J. R.; Moran, I.; Cnop, M.; Torrents, D.

2025-07-24 genetic and genomic medicine
10.1101/2025.07.24.25332120 medRxiv
Show abstract

Type 2 diabetes (T2D) is a complex and clinically heterogeneous disease. Although clustering approaches have defined clinical subtypes, their genetic and molecular architectures remain poorly understood. Efforts using genome-wide association studies (GWAS) have been constrained by modest sample sizes and conservative significance thresholds, limiting subgroup resolution. Leveraging recent expansions in cohort scale and diversity, we developed a new analytical framework that integrates GWAS with Machine Learning and eXplainable Artificial Intelligence. Applied to the UK Biobank, this strategy enabled the first comprehensive genetic and molecular characterization of T2D subtypes, identifying 184 genes that define each subgroups molecular landscape and associated tissue profiles. In addition, the framework improved detection of T2D risk compared to current Polygenic Risk Scores. The findings underscore the importance of clinical stratification to uncover the complex pathophysiology of T2D and to pave the way for more precise prevention and treatment strategies.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.