Genome-wide maps of transcription factor footprints identify noncoding variants rewiring gene regulatory networks with varTFBridge

Common-variant genome-wide association studies have identified thousands of noncoding loci associated with human diseases and complex traits; however, interpreting their functional mechanisms remains a major challenge. In recent years, the decreasing cost of high-throughput sequencing has enabled large population cohorts to generate whole-genome sequencing data for hundreds of thousands of individuals, providing an opportunity to comprehensively explore the functional impact of the full spectrum of genetic variants. However, unlike common-variant GWAS or rare-variant burden testing in coding regions, robust and systematic strategies for interpreting rare variants in noncoding regions remain limited. Meanwhile, many causal variants in non-coding regions are thought to act by perturbing transcription factor (TF) binding and rewiring gene regulatory networks, but progress has been limited by the lack of accurate, high-resolution maps of TF footprints. Recent single-molecule deaminase footprinting (FOODIE) technologies enable precise genome-wide detection of TF footprints at near-single-base resolution. We discovered that K562 FOODIE footprints, although comprising less than 0.5% of the genome, exhibit approximately 70-fold enrichment for erythroid trait heritability, establishing an unprecedented resource for interrogating TF-mediated regulatory mechanisms. We further introduce varTFBridge, an integrative framework that combines both common and rare non-coding variant association analyses, footprint-gene linking models, and AlphaGenome to prioritize causal noncoding variants that rewire gene regulatory networks. Across 13 erythrocyte traits, varTFBridge linked 209 credible common and 18 driver rare variants altering TF binding affinity to modulate 207 unique target genes. We successfully recapitulated a known causal variant associated with erythrocyte traits and further elucidated its molecular mechanism, revealing how specific disruption of TF co-binding alters CCND3 regulation to drive variation in red blood cell count. Together, our approach enables genome-wide functional predictions of trait-associated noncoding variants by linking transcription factor binding events to their target genes.