High Resolution Models of Transcription Factor-DNA Affinities Improve In Vitro and In Vivo Binding Predictions
Accurately modeling the DNA sequence preferences of transcription factors (TFs), and using these models to predict in vivo genomic binding sites for TFs, are key pieces in deciphering the regulatory code. These efforts have been frustrated by the limited availability and accuracy of TF binding site motifs, usually represented as position-specific scoring matrices (PSSMs), which may match large numbers of sites and produce an unreliable list of target genes. Recently, protein binding microarray (PBM) experiments have emerged as a new source of high resolution data on in vitro TF binding specificities. PBM data has been analyzed either by estimating PSSMs or via rank statistics on probe intensities, so that individual sequence patterns are assigned enrichment scores (E-scores). This representation is informative but unwieldy because every TF is assigned a list of thousands of scored sequence patterns. Meanwhile, high-resolution in vivo TF occupancy data from ChIP-seq experiments is also increasingly available. We have developed a flexible discriminative framework for learning TF binding preferences from high resolution in vitro and in vivo data. We first trained support vector regression (SVR) models on PBM data to learn the mapping from probe sequences to binding intensities. We used a novel -mer based string kernel called the di-mismatch kernel to represent probe sequence similarities. The SVR models are more compact than E-scores, more expressive than PSSMs, and can be readily used to scan genomics regions to predict in vivo occupancy. Using a large data set of yeast and mouse TFs, we found that our SVR models can better predict probe intensity than the E-score method or PBM-derived PSSMs. Moreover, by using SVRs to score yeast, mouse, and human genomic regions, we were better able to predict genomic occupancy as measured by ChIP-chip and ChIP-seq experiments. Finally, we found that by training kernel-based models directly on ChIP-seq data, we greatly improved in vivo occupancy prediction, and by comparing a TF's in vitro and in vivo models, we could identify cofactors and disambiguate direct and indirect binding. Transcription factors (TFs) are proteins that bind sites in the non-coding DNA and regulate the expression of targeted genes. Being able to predict the genome-wide binding locations of TFs is an important step in deciphering gene regulatory networks. Historically, there was very limited experimental data on the DNA-binding preferences of most TFs. Computational biologists used known sites to estimate simple binding site motifs, called position-specific scoring matrices, and scan the genome for additional potential binding locations, but this approach often led to many false positive predictions. Here we introduce a machine learning approach to leverage new high resolution data on the binding preferences of TFs, namely, protein binding microarray (PBM) experiments which measure the in vitro binding affinities of TFs with respect to an array of double-stranded DNA probes, and chromatin immunoprecipitation experiments followed by next generation sequencing (ChIP-seq) which measure in vivo genome-wide binding of TFs in a given cell type. We show that by training statistical models on high resolution PBM and ChIP-seq data, we can more accurately represent the subtle DNA binding preferences of TFs and predict their genome-wide binding locations. These results will enable advances in the computational analysis of transcriptional regulation in mammalian genomes.