Automated incorporation of pairwise dependency in transcription factor binding site prediction using dinucleotide weight tensors

Abstract

Gene regulatory networks are ultimately encoded by the sequence-specific binding of (TFs) to short DNA segments. Although it is customary to represent the binding specificity of a TF by a position-specific weight matrix (PSWM), which assumes each position within a site contributes independently to the overall binding affinity, evidence has been accumulating that there can be significant dependencies between positions. Unfortunately, methodological challenges have so far hindered the development of a practical and generally-accepted extension of the PSWM model. On the one hand, simple models that only consider dependencies between nearest-neighbor positions are easy to use in practice, but fail to account for the distal dependencies that are observed in the data. On the other hand, models that allow for arbitrary dependencies are prone to overfitting, requiring regularization schemes that are difficult to use in practice for non-experts. Here we present a new regulatory motif model, called dinucleotide weight tensor (DWT), that incorporates arbitrary pairwise dependencies between positions in binding sites, rigorously from first principles, and free from tunable parameters. We demonstrate the power of the method on a large set of ChIP-seq data-sets, showing that DWTs outperform both PSWMs and motif models that only incorporate nearest-neighbor dependencies. We also demonstrate that DWTs outperform two previously proposed methods. Finally, we show that DWTs inferred from ChIP-seq data also outperform PSWMs on HT-SELEX data for the same TF, suggesting that DWTs capture inherent biophysical properties of the interactions between the DNA binding domains of TFs and their binding sites. We make a suite of DWT tools available at dwt.unibas.ch, that allow users to automatically perform ‘motif finding’, i.e. the inference of DWT motifs from a set of sequences, binding site prediction with DWTs, and visualization of DWT ‘dilogo’ motifs.

JASPAR 2018: update of the open-access database of transcription factor binding profiles and its web framework

“…An intrinsic limitation to PFMs/PWMs is that they ignore inter-nucleotide dependencies within TFBSs ( 9 – 13 ). TF–DNA interaction data derived from next-generation sequencing assays has improved the computational modeling of TF binding ( 14 – 19 ). For example, the TF flexible models (TFFMs) ( 14 ), based on first-order hidden Markov models, capture dinucleotide dependencies within TFBSs and were introduced in the 2016 release of the JASPAR database.…”

Position-specific evolution in transcription factor binding sites, and a fast likelihood calculation for the F81 model

“…It is widely recognized that PWMs are not adequate representations of TFBS complexity, which contain significant positional correlations [18] and multiple efforts have been made to go beyond the PWM model [2][3][4][5]. As a complementary approach, we show that positional dependency effects can be examined by conditioning PSSVs on specific ancestral nucleotides.…”

“…However, positional dependencies are known to occur in binding sites of many TFs. Extensions to the basic PWM format have been proposed to account for positional dependencies [2][3][4][5]. Nevertheless, PWMs persist as the dominant motif representation format, because of ease of calculation and because they are easily visualized via sequence logos [6].…”

“…8,12,18,2,15,17,6,4,11,5,16,19,14,13,9,1,3,10,7),(11,5,4,9,2,16,10,12,14,7,6,8,18,19,3,13,15,17,1) and(13,16,9,5,7,6,4,11,1,2,12,17,18,8,10,15,…”

erefinder: Genome‐wide detection of oestrogen response elements

“…Position weight matrices assume each substitution at a base pair position has an independent effect on the binding affinity of the protein to the motif and the magnitude of the effect is related to conservation of the base pair position in the frequency matrix (Stormo, 2000). There is no shortage of programs that can search sequences based on the position weight matrix (Frith, Li, & Weng, 2003;Kel et al, 2003;Tan & Lenhard, 2016;Wang, Martins, & Danko, 2016); however, the generation of a position frequency matrix can involve bias (Teytelman, Thurtle, Rine, & van Oudenaarden, 2013) and the assumption of independence of base pairs may often be unwarranted (Bulyk, Johnson, & Church, 2002;Man & Stormo, 2001;Omidi et al, 2017). The lack of independence is especially nontrivial in EREs, as loss of a perfect half site has a larger effect on the binding affinity than point mutations after the half site is lost (Deegan et al, 2011;Tyulmenkov & Klinge, 2001).…”