PIQ is a new computational method that uses machine learning to identify the
genomic binding sites of hundreds of transcription factors (TFs) at
corresponding motifs from DNase-Seq experiments with accuracy comparable to ChIP-Seq.
We employed PIQ in a developmental lineage paradigm of pancreatic endoderm
specification from embryonic stem cells and identified a set of novel pioneer TFs
that dynamically open chromatin in the lineage and enable other TFs to bind to adjacent DNA.
Surprisingly, several pioneer TFs with non-palindromic motifs only open chromatin
in one direction. We found that genomic binding of a distinct subset of settler TFs
depends on proximity to chromatin opened by pioneer TFs. In vivo experiments
confirmed directional and non-directional pioneer activity and settler binding.