Based on HT-SELEX data from Yin et. al., we trained deltaSVM models for 533 distinct transcription factors(TFs) that could predict differential TF binding to SNPs. We then evaluated the performance of 129 out of 533 models using SNP-SELEX data and selected 94 models with high accuracy (AUPRC>0.75) for genome-wide prediction.

Here, we provide models to make predictions for differential binding to SNPs of those 94 TFs. For few SNPs of interest (< 1,000), our web server GVATdb is recommended. To predict a large set of SNPs, you can clone GitHub Repo and run it on a local server.