SRA

A large number of sequence variants have been linked to complex human traits and diseases, but deciphering their biological functions is still challenging since most of them reside in the noncoding DNA. To fill this gap, we have systematically assessed the binding of 270 human transcription factors (TF) to 95,886 noncoding variants in the human genome using an ultra-high-throughput multiplex protein-DNA binding assay, termed SNP evaluation by Systematic Evolution of Ligands by EXponential enrichment (SNP-SELEX). The resulting 828 million measurements of TF-DNA interactions enable estimation of the relative affinity of these TFs to each variant in vitro and allow for evaluation of the current methods to predict the impact of noncoding variants on TF binding. We show that the Position Weight Matrices (PWMs) of most TFs lack sufficient predictive power, while the Support Vector Machine (SVM) combined with the gapped k-mer representation show much improved performance, when assessed on results from independent SNP-SELEX experiments involving a new set of 61,020 sequence variants. We report highly predictive models for 94 human TFs and demonstrate their utility in genome-wide association studies (GWAS) and understanding of the molecular pathways involved in diverse human traits and diseases. Overall design: 768 experiments with HT-SELEX for six SELEX cycles to measure allelic TF binding for 95,886 SNPs. TF ChIP-seq and RNA-seq were performed in HepG2 cells to validate allelic TF binding. STARR-seq experiments were performed to identify SNPs that affect enhancer activity in HepG2 and HEK293 cells with three replicates. In situ Hi-C experiments were performed in HepG2 cells and human islets to identify target genes of SNPs.