VarNexus: an integrative database for interpreting disease-associated regulatory variants through multimodal predictions and functional evidence

Abstract

Genome-wide association studies (GWAS) have identified numerous loci associated with complex diseases and traits, yet the functional interpretation of non-coding risk variants remains challenging. Sequence-to-function models like AlphaGenome provide large-scale multi-modal predictions of variant effects, but their predictions lack direct experimental support. In contrast, empirical resources such as molecular quantitative trait loci (QTLs) and massively parallel reporter assays (MPRAs) offer population-based or experimental support for variant function, but suffer from limited genomic coverage. Here, we present VarNexus (https://biomed.imu.edu.cn/varnexus), an integrative resource for connecting disease-associated variants with predicted, population-based, and experimental evidence. VarNexus currently contains 230 404 fine-mapped candidate variants across 1891 disease-related phenotypes from UK Biobank and FinnGen. It integrates precomputed AlphaGenome predictions across 11 regulatory output types with bulk and single-cell xQTLs, MPRA measurements, and GWAS-eQTL colocalization results. VarNexus enables cross-linked searches by variant, phenotype, and gene, together with on-demand AlphaGenome prediction, multi-omics genome-browser visualization, and regulatory network exploration. By combining sequence-based predictions with population-based and experimental evidence, VarNexus provides a framework for prioritizing candidate regulatory mechanisms supported by multiple evidence layers.