SeqGL is a new group lasso-based algorithm to extract multiple transcription factor (TF) binding signals from ChIP- and DNase-seq profiles. Benchmarked on over 100 ChIP-seq experiments, SeqGL outperformed traditional motif discovery tools in discriminative accuracy and cofactor detection. SeqGL successfully scales to DNase-seq data, identifying a large multiplicity of TF signals confirmed by ChIP, and can be used with multitask training to learn genomic-context and cell-type specific TF signals.
This repository is an R package to run SeqGL. Please refer to the vignette for installation and usage instructions. This package has been tested with R 3.2. The package and the vignette are available to download from the "Downloads" tab.
SeqGL identifies context-dependent binding signals in ChIP-seq and DNase-seq profiles. Manu Setty & Christina S Leslie. PLoS Computational Biology. May 2015. Link