PeakRegressor Identifies Composite Sequence Motifs Responsible for STAT1 Binding Sites and Their Potential rSNPs

Abstract

How to identify true transcription factor binding sites on the basis of sequence motif information (e.g., motif pattern, location, combination, etc.) is an important question in bioinformatics. We present “PeakRegressor,” a system that identifies binding motifs by combining DNA-sequence data and ChIP-Seq data. PeakRegressor uses L1-norm log linear regression in order to predict peak values from binding motif candidates. Our approach successfully predicts the peak values of STAT1 and RNA Polymerase II with correlation coefficients as high as 0.65 and 0.66, respectively. Using PeakRegressor, we could identify composite motifs for STAT1, as well as potential regulatory SNPs (rSNPs) involved in the regulation of transcription levels of neighboring genes. In addition, we show that among five regression methods, L1-norm log linear regression achieves the best performance with respect to binding motif identification, biological interpretability and computational efficiency.

Discriminative motif analysis of high-throughput dataset

“…Our software also identifies depleted motifs, but they were ignored here. To find the annotated motif of the ChIP-ed TF, we matched TF names/aliases with the motif names in the motif databases Jaspar (Bryne et al, 2008;Redhead and Bailey, 2007) and Uniprobe (Newburger and Bulyk, 2009). If no exact matches were found, we used the motif of a homolog; e.g.…”

“…Discriminative motif discovery is not a new approach. Pioneering work includes, but is not limited to, DME (Smith et al, 2006), DIPS (Sinha, 2006) and DEME (Redhead and Bailey, 2007). These methods find a discriminative position weight matrix (PWM) to optimize an objective function, which for the case of DEME and DME, is the likelihood of the data given the model and sequence class.…”

“…Recent works designed for high-throughput datasets use more simplified statistical models. For example, DREME [MEME suite (Bailey, 2011)] and oligo-diff [RSAT suite (Thomas-Chollier et al, 2012)] use Fisher's exact test and PeakRegressor (Pessiot et al, 2010) applies a linear regression model to fit peak scores by motif counts.…”