CrobustaTFs

An R package containing candidate Ciona robusta motifs developed for use with ATAC-seq data. It integrates motifs identified using SELEX-seq with known orthologs from HOMER and inferred orthologs from [Cis-BP](doi: 10.1016/j.cell.2014.08.009). The install script attempts to retrieve all source files and runs a Gale-Shapley-like algorithm to map candidate motifs to genes with minimal ambiguity

Rationale

Motifs in C. robusta are much less well characterized than H. sapiens or M. musculus. The most common techniques to fill in the gaps use either orthologs from other chordates or to infer orthologs from protein DNA binding domains. These methods naturally raise the question of which motif is "best." The answer often comes down to experimenter judgment, and different sources may yield more useful motifs for different TFs. The problem is particularly pronounced with inferred orthologs, as there are often many-to-many relationships between motifs and TFs. It is often desirable to be able to unambiguously assign a motif to a singe TF. This is done by iteratively removing associations of ambiguous motifs to TFs that are already unambiguously associated to a motif. This package aims to simplify mapping motifs to TFs in C. robusta by unifying possible sources into a single database.

Installation

Depends: R (>= 2.10), TFBSTools

Imports: DBI, motifmatchr

git clone https://github.com/kewiechecki/CrobustaTFs
cd CrobustaMotifs
make

Usage

library(CrobustaMotifs)

# accessing tags(CrobustaMotifs) requires loading the TFBSTools library separately
library(TFBSTools)

# function to compress motif metadata into a list of data.frames
tfTags <- function(motifs,tag){
	res <- setNames(do.call(data.frame,
		lapply(tag,
		       function(x) sapply(tags(motifs),'[[',x))),
		tag)
	return(res)
}

# extract fields from tags(CrobustaMotifs)
meta <- tfTags(CrobustaMotifs, 
	       names(tags(CrobustaMotifs[[1]])))

# split metadata by TF
meta <- split(meta,meta$KYID)

# select lowest mean shannon entropy motif for each TF
sel <- sapply(meta,function(x) row.names(x)[which.min(x$meanEntropy)])
motifsub <- CrobustaMotifs[sel]