R/originalCode/construct_summary.8merscores_ref_SS.v2.R

In alice-szeliga/keles-pipeline: A Model to Predict Regulatory and Expression Modulating eQTLs

#############################
rm(list=ls())
library(data.table)
load("/p/keles/CAGI2015/volumeB/ProcessedData/escores8mer_ref_human_v2.Rda")
enrscoresList_human<-enrscoresList
NameM_human<-NameM

load("/p/keles/CAGI2015/volumeB/ProcessedData/escores8mer_ref_mouse_v2.Rda")
enrscoresList_mouse<-enrscoresList
NameM_mouse<-NameM

enrscoresList<-c(enrscoresList_human, enrscoresList_mouse)
NameM<-rbind(NameM_human, NameM_mouse)

#####Add unique protein names#####
ProteinNames<-NameM[,2]
ProteinNames[grep("Ehf", NameM[,2])]<-c("Ehf_Wei", "Ehf_Badis")
ProteinNames[grep("Elf3", NameM[,2])]<-c("Elf3_Wei", "Elf3_Badis")
ProteinNames[grep("Gabpa", NameM[,2])]<-c("Gabpa_Wei", "Gabpa_Badis")
ProteinNames[grep("Hoxa3", NameM[,2])]<-c("Hoxa3_Badis", "Hoxa3_Berger")
ProteinNames[grep("Nkx3-1", NameM[,2])]<-c("Nkx3-1_Badis", "Nkx3-1_Berger")
ProteinNames[grep("Sfpi1", NameM[,2])]<-c("Sfpi1_Wei", "Sfpi1_Badis")
ProteinNames[grep("Six6", NameM[,2])]<-c("Six6_Badis", "Six6_Berger")
ProteinNames[grep("Sox4", NameM[,2])]<-c("Sox4_Human", "Sox4_Mouse")
ProteinNames[grep("Spdef", NameM[,2])]<-c("Spdef_Wei", "Spdef_Badis")
ProteinNames[grep("Tcf1", NameM[,2])]<-c("Tcf1_Badis", "Tcf1_Berger")

ProteinNames[grep("GST-CSL", NameM[,2])[1:2]]<-c("GST-CSL_Fig2", "GST-CSL_Fig4")
ProteinNames[grep("MAML1-CSL-GST-NOTCH1", NameM[,2])]<-c("MAML1-CSL-GST-NOTCH1_Fig2", "MAML1-CSL-GST-NOTCH1_Fig5")
ProteinNames[grep("MAML1-CSL-GST-NOTCH2", NameM[,2])]<-c("MAML1-CSL-GST-NOTCH2_Fig5", "MAML1-CSL-GST-NOTCH2_FigS1")

ProteinNames[grep("FOXN4", NameM[,2])]<-"FOXN4_Human"
ProteinNames[grep("Foxn4", NameM[,2])]<-"Foxn4_Mouse"
ProteinNames[grep("GST-NOTCH1", NameM[,2])[1:4]]<-c("GST-NOTCH1_CSL-6His", "GST-NOTCH1_CSL-6His_MAML1", "GST-NOTCH1_Fig2", "GST-NOTCH1_Fig4")

names(enrscoresList)<-ProteinNames
##############

SumEscores<-mapply(function(enrList) mapply(sum, enrList), enrscoresList)
MeanEscores<-mapply(function(enrList) mapply(mean, enrList), enrscoresList)
MaxEscores<-mapply(function(enrList) mapply(max, enrList), enrscoresList)

load("/p/keles/CAGI2015/volumeB/ProcessedData/SampleDT.RData")

lenCenter<-mapply(nchar, trainingDT$RefAllele)
table(trainingDT$pos-trainingDT$start+1)
table(lenCenter)

GetSubseqCoordAbs<-function(subseq.len=21, center.len=lenCenter, center.pos=trainingDT$pos) {
heading<-ceiling((subseq.len-center.len)/2)
foot<-floor((subseq.len-center.len)/2)
start<-center.pos-heading
end<-center.pos+center.len-1+foot
subseq.coord<-cbind(start, end)
subseq.coord
}

subseq.coord.abs<-GetSubseqCoordAbs(subseq.len=21, center.len=lenCenter, center.pos=trainingDT$pos)
subseq.coord.rel<-subseq.coord.abs-matrix(trainingDT$start, nrow(trainingDT), 2)+1
subseq.coord.rel.list<-lapply(seq_len(nrow(subseq.coord.rel)), function(i) subseq.coord.rel[i,])

enrscoresSubseq21List<-mapply(function(enrscores) t(mapply(function(enrscore, rel.pos) enrscore[rel.pos[1]:rel.pos[2]], enrscores, subseq.coord.rel.list)), enrscoresList, SIMPLIFY=FALSE)

SumSubseq21Escores<-mapply(rowSums, enrscoresSubseq21List)
MeanSubseq21Escores<-mapply(rowMeans, enrscoresSubseq21List)
MaxSubseq21Escores<-mapply(function(enrM) apply(enrM, 1, max), enrscoresSubseq21List)

escoresList<-vector("list")
escoresList[[1]]<-SumEscores; escoresList[[2]]<-MeanEscores; escoresList[[3]]<-MaxEscores
escoresList[[4]]<-SumSubseq21Escores; escoresList[[5]]<-MeanSubseq21Escores; escoresList[[6]]<-MaxSubseq21Escores
names(escoresList)<-c("sum", "mean", "max", "sum_bp21", "mean_bp21", "max_bp21")


save(escoresList, file="/p/keles/CAGI2015/volumeB/ProcessedData/Features8mersEscores_ref_v2.Rda")