Integrating machine learning and GWAS for variant prioritization in the INCIPE cohort highlights ABC transporter genes in chronic kidney disease.
- Open access
Machine learning model achieved 87.77% ROC AUC, significantly enhancing variant prioritization in CKD within a small, imbalanced cohort.
- Why it matters: Early detection of CKD relies on identifying genetic biomarkers, but traditional GWAS struggles with limited power and false results in small samples, hindering progress.
- What they did: Researchers used a nested ensemble machine learning approach combining undersampling and CatBoostClassifier to prioritize genetic variants, followed by functional and expression analyses, validating findings with external CKD gene sets.
- The result: The model identified candidate genes, notably ABC transporter genes like ABCA13, ABCA4, and ABCC4, with ABCA4 highly expressed in kidney tissue, advancing understanding of CKD genetics and potential biomarkers.