Base-resolution methylation patterns accurately predict transcription factor bindings in vivo
Nucleic Acids Res. 2015 Mar 11;43(5):2757-66. doi: 10.1093/nar/gkv151. Epub 2015 Feb 26.
Methylphet adopts a two-step method to predict transcription factor-DNA interaction using DNA methylation profiles from whole-genome bisulfite sequencing data by exploiting the connection between DNA methylation level and transcription factor binding. In the first step, beta-binomial models are devised to characterize DNA methylation data around TF binding sites and the background to estimate methylation scores. Along with other static genomic features, a random forest framework is adopted in the second step to predict transcription factor-DNA interaction. When all methylation profile are taken together and combined with features at the sequence level, Methylphet can accurately predict TF binding and performs favorably when compared against competing methods.
Package "devtools" is required to install this package from Github
After installing devtools, use the code below to install Methylphet:
library("devtools") install_github("Methylphet", username="StanleyXu”) ?Methylphet
### Load the package library("Methylphet") ### Load all the data needed data(list = data(package="Methylphet")$results[,3]) ### Using CpG methylation information only to predict TFBS when 0/1 golden standard is provided. OCT4_CpG = Methylphet(traindata.mat = mESdata.motif.chr10,traindata.methyl1=mESdata.CpG.chr10, goldstandard=goldstandard.chr10, OtherGenomicFeatures.train=OtherGenomicFeatures.mES.chr10, testdata.mat =H1data.motif.chr10, testdata.methyl1=H1data.CpG.chr10, OtherGenomicFeatures.test=OtherGenomicFeatures.H1.chr10) ### Using both CpG and CpH methylation information to predict TFBS when location for ChIP-seq peaks are provided. OCT4_CpG_CpH = Methylphet(traindata.mat = mESdata.motif.chr10, traindata.methyl1=mESdata.CpG.chr10,traindata.methyl2=mESdata.CpH.chr10, ChIPseqPeaks = peak.GR.ES.chr10, OtherGenomicFeatures.train=OtherGenomicFeatures.mES.chr10, testdata.mat =H1data.motif.chr10, testdata.methyl1=H1data.CpG.chr10,testdata.methyl2=H1data.CpH.chr10, OtherGenomicFeatures.test=OtherGenomicFeatures.H1.chr10)