We have developed and compared different methods for predicting TF target genes using ChIP-seq data.We implemented all the methods that we tested in a unified R-package, thus providing a convenient tool for testing different methods on your own data. Here you can download our R-package TargetCaller.
Link to the paper describing the target prediction methods.The current version contains code to compute FDR-corrected q-values via permutations of the peak-to-gene assignments (ClosestGene only). Check the file DESCRIPTION and the R-documentation for details how to use the code.
We used the following ChIP-seq data for evaluating the methods:
Gene expression data used for the validation:
perturbationList.RData (3 MB)
This file contains an R-object with a ‘perturbationList’ object. This is a list of 21 elements corresponding to the selected TFs from Table S1. Each element is a named vector of fold change values prepared as described in the paper. These vectors contain all the genes in the experiment. For the paper the top 500 genes with the largest absolute fold changes were used.
Beyer A, Workman C, Hollunder J, Radke D, Möller U, Wilhelm T, Ideker TG (2006) Integrated assessment and prediction of transcription factor binding. PLoS Comput. Biol. 2(6):e70.
Ucar D, Beyer A, Parthasarathy S, Workman C (2008) Predicting functionality of protein-DNA interactions by integrating diverse evidence. Bioinformatics. 25(12):i137-44.
Sikora-Wohlfeld W, Ackermann M, Christodoulou EG, Singaravelu K, Beyer A (2013) Assessing Computational Methods for Transcription Factor Target Gene Identification Based on ChIP-seq Data. PLoS Comput. Biol. 9(11): e1003342.