R/OverlapAnalysisFunctions.R
In biodev/packageDir: Tool for in silico functional analysis design and exploration of relation between genomic function and aberration.
Defines functions
RunOverlapAnalysis
test.abDrugOverlapAnalysis
abDrugOverlapAnalysis
selectOverlapType
runUnmatchedCohort
pickSettings
runMatchedCohort
runMatchedIndividuals
runUnmatchedIndividuals
CheckPatientOverlap
plotOverlapArea
adjustOverlapToSensitive
extractedDrugScreenAnalysis
combineAbSens
coreOverlapAnalysis2
coreOverlapAnalysis
nullOrEmpty
Documented in
abDrugOverlapAnalysis
coreOverlapAnalysis
#overlapAnalaysisFunctions
RunOverlapAnalysis<-function(settings, study){
results = abDrugOverlapAnalysis(study=study,
settings=settings)
return(results)
}
test.abDrugOverlapAnalysis<-function(){
#stud = STUDY
#test all instances with individuals and cohorts.
#test matched cohort
#get default overlap analysis settings
defSettings = getTestStudySettings()
overlapSettings = defSettings$overlap_analysis
#get a study object pre-loaded with data aberration and functional data.
stud = getTestStudyObject()
#run the overlap analysis with functional genomic and genomic aberration data from the same cohort
overlapRes1 = abDrugOverlapAnalysis(study=stud, settings=overlapSettings)
#simulate the analysis with functional genomic and genomic aberration data from sepparate cohorts
studySepparateCohort = stud
patientSums = studySepparateCohort@results$functional_drug_screen_summary$patientsums
rownames(patientSums) <- toupper(rownames(patientSums))
studySepparateCohort@results$functional_drug_screen_summary$patientsums = patientSums
overlapRes2 = abDrugOverlapAnalysis(study=studySepparateCohort,
settings=overlapSettings)
}
#settings slots expected:
# combinedAberrations
#'@title abDrugOverlapAnalysis
#'@description main function for running analysis of functional and genomic aberration arms
#'@param study A study object (must have results loaded for aberration and functional analyses)
#'@param thresh Numeric, the threshold used to establish which pathways will be considered to be drug sensitive and aberrational.
#'@param enrich_col String, the name of the column in the path summary used to determine significance of a pathway. This is the column to which the threshold argument is to be applied.
#'@param verbose Logical, a flag indicating if the program is to provide extra diagnostic output.
#'@param settings List obejct. This contains the settings used in the overlap analysis. Use getDefaultOverlapSettings()
#'@return Study object with overlap analysis in results list added or replaced
#'@export
#'@examples
#'#get a study object pre-loaded with genomic aberration and functional genomic data.
#' stud = getTestStudyObject()
#'
#'#get default overlap analysis settings
#'defSettings = getTestStudySettings()
#'overlapSettings = defSettings$overlap_analysis
#'
#'#run the overlap analysis
#'
#'overlapRes1 = abDrugOverlapAnalysis(study=stud, settings=overlapSettings)
#'
#'\dontrun{
#'#If settings for the overlap analysis cannot be found, the overlap
#'#analysis will run in interactive mode, requiring user input.
#'#This line of code will run the overlap anaysis in interactive mode,
#'#requesting user input to establish settings:
#'
#'abDrugOverlapAnalysis(study=stud)
#'}
#'@importFrom VennDiagram venn.diagram
#'@importFrom calibrate textxy
abDrugOverlapAnalysis<-function(study,
thresh=0.05,
enrich_col="hyperg_p_w_FDR",
verbose=T,
settings=NULL){
if(!exists("VERBOSE")) VERBOSE=T
cat("\n******************************************************************************************\n")
cat("\n*************************************overlap analysis*************************************\n")
cat("\n******************************************************************************************\n")
checkPathsMatch(STUDY=study)
sfolder = paste(studyFolder(s=study),"/results/", sep="")
#set up ola object
ola = list() #ola = overlap analysis
ola$results = study@results
ola$path_detail = getPaths(study)
ola$threshold = thresh
ola$enrich_col = enrich_col
ola$outfname_path = sfolder
ola$verbose = verbose
if("interactive"%in%names(settings)){
#if interactive is found as a flag in settings it is the first time overlap analysis has been run for the study
#put it in the right place
if( "basic_overlap"%in%names(settings) ){#overlap analysis has already been run
settings_full = settings #transfer the settings list
settings_full$interactive = NULL #erase the flag from the master list
interactiveFlag = settings$interactive # put the interactive flag in the right place
settings = settings_full$basic_overlap # in the
settings$interactive = interactiveFlag
}else{#it's the first time the overlap was run on the current study
settings_full = list()
settings_full$basic_overlap = settings
}
}else{
settings_full = settings
settings = settings_full$basic_overlap
}
# settings$interactive = settings_full$interactive
# settings_full$interactive = NULL
settings = setting(s=settings, prompt="Would you like to run the function/aberration overlap analysis on each patient? (y/n)")
runEachPatient=settings$.text == "y"
ola$runEachPatient = runEachPatient
while(T){
settings = setting(s=settings, "Would you like to use default path analysis settings or use those associated with study arms? (please enter d or a)")
if(settings$.text%in%c("a","d")) break
print("Sorry, your input was not understood, please try again.")
}
if( settings$.text=="d" ){
settings_full$defaultSummaryTable = getDefaultSettings()$defaultSummaryTable
settings$interactive = F
}else{
settings_full$defaultSummaryTable = NULL
settings$interactive = T
}
settings_full$basic_overlap = settings
ola$settings = settings_full
ola = selectOverlapType(ola=ola, study=study)
#system('/usr/bin/afplay ./reference_data/Submarine.aiff')
ola$results$overlap_analysis$settings = ola$settings
return(ola$results$overlap_analysis)
}
selectOverlapType<-function(ola, study){
#determine if matched
overlapPatients = CheckPatientOverlap(results=ola$results)
#Assure only the sensitive patients are
overlapPatients = adjustOverlapToSensitive(overlap=overlapPatients,
functionalEnrichmentAnalysis=ola$results$functional_drug_screen_summary)
matched = length(overlapPatients)>0 #if there are overlap patients, it's assumed to be a matched sample
if( matched ){
#1: cohort, matched
#limit cohort
ola = runMatchedCohort(overlapPatients=overlapPatients, ola=ola, study=study)
}else{
cat("\nNote: no patient sample ids were found to be associated with\nboth sensitive gene targets and aberration analysis/analyses.\n")
#2: cohort, unmatched
#no limiting, just run it once the combination is done
ola = runUnmatchedCohort(ola=ola, study=study)
}
if( ola$runEachPatient ){#if the overlap analysis should be run for each patient
ola$outfname_path = paste(ola$outfname_path, "overlap_analysis_each_patient/", sep="")
if(matched){
#3: individual patient, matched
#pass drug screen summary for patient plus aberration summary for patient
ola = runMatchedIndividuals(overlapPatients=overlapPatients, ola=ola, study=study)
}else{
#4: individual patient, unmatched
#pass drug screen summary for individual patient, plus aberration summary for whole cohort
ola = runUnmatchedIndividuals(ola=ola, study=study)
}
#indresults should be the list with each slot named with a patient identifier
}
return(ola)
}
runUnmatchedCohort<-function(ola, study, functional_analysis_name="functional_drug_screen_summary"){
if(VERBOSE) cat("\nola$settings:\n")
if(VERBOSE) print(ola$settings)
functionalEnrichmentAnalysis = ola$results[[functional_analysis_name[1]]]
#case 1: use what was passed as settings
# combSettings = ola$settings$combinedAberrations
if(is.null(ola$settings$defaultSummaryTable)){#if the default settings aren't provided, look for the settings in the arms
ola$settings$combinedAberrations = NULL
#case 2: use the default settings
#case 3: use the settings from an aberration arm (taken care of in combineAberrationTypes)
# ---> if the user said not to use the default, the combineAberrationTypes2 will make the user pick
}else{
ola$settings$combinedAberrations = ola$settings$defaultSummaryTable
}
if(is.null(ola$results$combinedAberrations)) ola$results$combinedAberrations = combineAberrationTypes(study=study,
s=ola$settings$combinedAberrations,
results=ola$results,
runEachPatient=F)
ola$settings$combinedAberrations = ola$results$combinedAberrations$settings
combinedAberrations = ola$results$combinedAberrations
ola$outfname_path = paste(ola$outfname_path, "overlap_analysis/", sep="")
ola$results$overlap_analysis = coreOverlapAnalysis(functionalEnrichmentAnalysis=functionalEnrichmentAnalysis,
combinedAberrations=combinedAberrations,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
return(ola)
}#runUnmatchedCohort
pickSettings<-function(study, settingType = "aberration"){
si = grep(pattern=settingType, x=names(study@studyMetaData@settings), ignore.case=T)
targetArmSettings = names(study@studyMetaData@settings)[si]
if(length(si)>1){
sres = settingList(s=list(interactive=T),
prompt="Please pick a study arm whose path analysis settings you would like to use.",
set=names(study@studyMetaData@settings)[si])
targetArmSettings = sres$.text
}
return(study@studyMetaData@settings[[targetArmSettings]])
}
runMatchedCohort<-function(ola, overlapPatients, study){
# if(is.null(ola$settings$combinedAberrations)) ola$settings$combinedAberrations =
#case 1: use the settings passed
funSettings = ola$settings$functionalPathAnalysis
if(is.null(ola$settings$defaultSummaryTable)){#if case 1 is null
ola$settings$functionalPathAnalysis = NULL
#case 2: use the default settings
#case 3: use the settings from an aberration arm (taken care of in combineAberrationTypes)
# ---> if the user said not to use the default, the combineAberrationTypes2 will make the user pick
}else{
ola$settings$functionalPathAnalysis = ola$settings$defaultSummaryTable
}
cat("\n\nAnalysis of functional drug screen for patients who have had both aberration and functional analysis..\n")
functionalEnrichmentAnalysis = extractedDrugScreenAnalysis(psubset=overlapPatients,
study=study,
s=ola$settings$functionalPathAnalysis,
results=ola$results,
path_detail=ola$path_detail,
eachPatient=ola$runEachPatient) #holds the results for the functional enrichment analysis
ola$results$overlap_analysis$functional_drug_screen_summary = functionalEnrichmentAnalysis
ola$settings$functionalPathAnalysis = functionalEnrichmentAnalysis$settings
# overlapPatients = intersect(overlapPatients, colnames(functionalEnrichmentAnalysis$patientsums))
if(is.null(ola$results$combinedAberrations)){
#case 1: use the settings passed
if(is.null(ola$settings$defaultSummaryTable)){#if the default settings aren't provided, look for the settings in the arms
ola$settings$combinedAberrations = NULL
#case 2: use the default settings
#case 3: use the settings from an aberration arm (taken care of in combineAberrationTypes)
# ---> if the user said not to use the default, the combineAberrationTypes2 will make the user pick
}else{
ola$settings$combinedAberrations = ola$settings$defaultSummaryTable
}
cat("\n\nAnalysis of combined aberration data for patients who have had both aberration and functional analysis..\n")
ola$results$combinedAberrations = combineAberrationTypes2(study=study,
s=ola$settings$combinedAberrations,
results=ola$results,
overlapPatients=overlapPatients)
}
ola$settings$combinedAberrations = ola$results$combinedAberrations$settings
ola$outfname_path = paste(ola$outfname_path, "overlap_analysis/", sep="")
ola$results$overlap_analysis = coreOverlapAnalysis(functionalEnrichmentAnalysis=functionalEnrichmentAnalysis,
combinedAberrations=ola$results$combinedAberrations,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
return(ola)
}
runMatchedIndividuals<-function(ola, overlapPatients, study){
# functionalEnrichmentAnalysis = NULL #holds the results for the functional enrichment analysis
# combinedAberrations = NULL #holds the results for the combined aberration enrichment analysis
pol = list() #patient overlap analyses
tmpAbs = ola$settings$combinedAberrations
#pull out the appropriate summaries
if(is.null(tmpAbs)){
tmpAbs = ola$results$combinedAberrations$settings
ola$settings$combinedAberrations = tmpAbs
}
tmpAbs$"To limit by count of patients with genes in examined state, enter count here:\n(Enter 1 to skip limiting by count)" = "1"
tmpAbs$'To limit by proportion of cohort with genes in examined state, enter minimum proportion here:\n(enter 0 for no limit): '="0"
tmpDs = ola$settings$functional_drug_screen_summary
if(is.null(tmpDs)){
tmpDs = ola$settings$defaultSummaryTable
# tmpDs = ola$results$functional_drug_screen_summary$settings
# if(class(tmpDs)!="list") tmpDs = dfToList(df=tmpDs)
# tmpDs$interactive = ola$settings$basic_overlap$interactive
ola$settings$functional_drug_screen_summary = tmpDs
}
tmpDs$"To limit by count of patients with genes in examined state, enter count here:\n(Enter 1 to skip limiting by count)" = "1"
tmpDs$'To limit by proportion of cohort with genes in examined state, enter minimum proportion here:\n(enter 0 for no limit): '="0"
fpathtmp = ola$outfname_path
for(p in overlapPatients){
print(p)
ola$outfname_path=paste(fpathtmp, p,"/overlap_analysis/", sep="")
combsum = combineAberrationTypes2(results=ola$results,
study=study,
s=tmpAbs,
overlapPatients=p)
functionalSum = extractedDrugScreenAnalysis(psubset=p,
eachPatient=F,
study=study,
s=tmpDs,
results=ola$results,
path_detail=ola$path_detail)
#run the overlap analysis
pres = coreOverlapAnalysis(functionalEnrichmentAnalysis=functionalSum,
combinedAberrations=combsum,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
#put the results in the right place
pol[[p]] = list()
pol[[p]][["overlap_analysis"]] = pres
}
ola$results$overlap_analysis$overlap_analysis_each_patient = pol
return(ola)
}#runMatchedIndividuals
runUnmatchedIndividuals<-function(ola, study){
functionalEnrichmentAnalysis = ola$results$overlap_analysis$functional_drug_screen_summary #holds the results for the functional enrichment analysis
combinedAberrations = ola$results$overlap_analysis$combined_aberrations_summary #holds the results for the combined aberration enrichment analysis
#get the set of individuals to be analyzed
patientSet = rownames(ola$results$functional_drug_screen_summary$patientsums)
ola$results$overlap_analysis$overlap_analysis_each_patient = list()
pol=list()
fpathtmp = ola$outfname_path
for(p in patientSet){
print(p)
ola$outfname_path=paste(fpathtmp, p,"/overlap_analysis/", sep="")
#extract patient record, extract patient drug results
patientFunctionalAnalysis = functionalEnrichmentAnalysis$path_summary_each_patient[[p]]
pres = coreOverlapAnalysis(functionalEnrichmentAnalysis=patientFunctionalAnalysis,
combinedAberrations=combinedAberrations,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
pol[[p]] = list()
pol[[p]][["overlap_analysis"]] = pres
}
ola$results$overlap_analysis$overlap_analysis_each_patient = pol
return(ola)
}#runUnmatchedIndividuals
CheckPatientOverlap<-function(results){
cat("..checking patients with functional and aberrational analysis..")
psetOut = c()
#check if analysis will be limited to patients who are in both cohorts, the drugs screen and the aberration sets
funIndex = grep("functional", x=names(results))
psetOut = rownames(results[[funIndex]]$patientsums)
abIndex = grep("aberration_", x=names(results))
for(i in abIndex){
psetOut = intersect(x=psetOut, y=rownames(results[[i]]$patientsums))
}
cat(" ..checked...")
return(psetOut)
}#CheckPatientOverlap
# load("./testsSavePlotOverlapData.rda", verbose=T)
# tabout_targAndAb = allInputs$tabout_targAndAb
# results = allInputs$results
# abandtargplotname = allInputs$abandtargplotname
plotOverlapArea<-function(tabout_targAndAb, results, abandtargplotname, autosave=T){
# readline("inside plot overlap area()")
# allInputs = list(tabout_targAndAb=tabout_targAndAb, results=results, abandtargplotname=abandtargplotname)
# save(allInputs, file="./testsSavePlotOverlapData.rda")
if(!nrow(tabout_targAndAb)){
print("There are no pathways in the overlap between drug targeted/drug sensitive and aberrational")
return()
}
#######################
#### plot overlap area
if(VERBOSE) print("Plotting overlap area.. ")
if(VERBOSE) print(colnames(tabout_targAndAb))
# tabout_targAndAb$path_id = gsub(pattern=" ",replacement="\n",x=tabout_targAndAb$path_id)
p1=ggplot(data=tabout_targAndAb, aes(x=proportion_of_cohort_w_aberrational_gene_in_path,
y=drug_targeted_genes,
size=testable_path_length.x,
label=path_id))+
geom_point(alpha=0.5)+
scale_size(range=c(5,30), guide = guide_legend(title = "Number of genes in path"))+
geom_text(size=2, vjust=1)+
xlab("Proportion of cohort with aberrational gene in path")+
ylab("Number of drug targeted genes")+
ggtitle("Proportion affected vs number of drug targets")+
theme_bw()
print(p1)
if(autosave){
pfname = plotFileName(abandtargplotname)
ggsave(plot=p1, width=10, height=6, filename=pfname)
}
# png(filename=abandtargplotname,width=800,height=500)
# maxpathlength = max(tabout_targAndAb[,3])
# cexvals = (tabout_targAndAb[,3]/maxpathlength)*10
#
# plot(cex=cexvals,
# ylim=c(0,1.2*max(tabout_targAndAb[,"drug_targeted_genes"])),#)),
# x=tabout_targAndAb[,"proportion_of_cohort_w_aberrational_gene_in_path"],
# y=tabout_targAndAb[,"drug_targeted_genes"],#"drug_targeted_genes"],
# main="Proportion affected vs number of drug targets.",
# xlab="Proportion of cohort with aberrational gene in path",
# ylab="Number of drug targets")
# textxy(tabout_targAndAb[,"proportion_of_cohort_w_aberrational_gene_in_path"],
# tabout_targAndAb[,"drug_targeted_genes"],
# labs=1:nrow(tabout_targAndAb))#,
# #cx=1.3)
# legend(x="topright",
# legend=paste("Size = path length (",paste(range(tabout_targAndAb[,3]), collapse=" to "),"nodes)\nNumber = path number in the \'enriched and targeted\' table"),
# cex=.8, pch=1)
# dev.off()
return(pfname)
}#plotOverlapArea
adjustOverlapToSensitive<-function(overlap, functionalEnrichmentAnalysis){
#get the total sensitive genes per patient
psums = functionalEnrichmentAnalysis$patientsums
#extract the patient ids for those patients with active genes
keepers = rownames(psums)[psums[,1]>0]
#subset the overlap patients to only those patients with sensitive genes
newOverlap = intersect(overlap, keepers)
return(newOverlap)
}#adjustOverlapToSensitive
extractedDrugScreenAnalysis<-function(psubset, results, path_detail, study, s, eachPatient){
#extractedDrugScreenAnalysis
#runs path analysis on drug screen patient subset
#returns drug screen results set with coverage analysis appended
#first get the functional analysis out of results
cat("...inside extractedDrugScreenAnalysis()...")
findex=grep(pattern="functional", names(results),ignore.case=T)
if(length(findex)>1) warning("More than one functional analysis was found in the input to the overlap analysis. (ex: more than one drug screen) This is currently a functionality of the program which has not been tested!")
if(VERBOSE) print(names(results))
if(VERBOSE) print(findex)
fa = results[[findex]]
pgm = as.matrix(fa$patientGeneMatrix[,psubset,drop=F]) #extract the subset of patients from the patient gene matrix
covset = rownames(fa$coverage_summary$genesummary)
# tm = getTargetMatrix(tgenes=covset, paths=path_detail$paths)
if(VERBOSE) print("summaryTable from extractedDrugScreenAnalysis")
cat("\ndim pgm going into summaryTable:", dim(pgm), "\n")
if(!is.null(pgm) & (nrow(pgm)>0)){
if(is.null(s)) s = fa$settings
summ = summaryTable(study=study,
coverageGeneDescription="targeted",
coverageDataSetDescription="Coverage of functional analysis platform",
individualEnrichment=eachPatient,
coverage=covset,
settings=s,
pgm=pgm,
activeGeneDescription="functionally_affected",
dataSetDescription="Functional path enrichment in patient subset for function-aberration overlap analysis")
}else{
print("There were no sensitive genes in this patient")
summ = list()
}
summ$coverage_summary = fa$coverage_summary
return(summ)
}#extractedDrugScreenAnalysis
combineAbSens<-function(functionalEnrichmentAnalysis, combinedAberrations, ola, coverage=NULL){
}
coreOverlapAnalysis2<-function(functionalEnrichmentAnalysis, combinedAberrations, ola, coverage=NULL){
### combine into single table:
# functional path enrichments
# aberration path enrichments
# functional coverage
# aberration data coverage
combtab = combineAbSens(functionalEnrichmentAnalysis=functionalEnrichmentAnalysis, combinedAberrations=combinedAberrations, ola=ola, coverage=coverage)
### determine what kind of overlap
# 1) ab + sensitivity + panel coverage
# 2) ab + panel coverage
### find which paths are in each part of the overlap
#panel coverage:
# 1) dark: ab, not covered
# 2) light: covered, not ab
# 3) prioritized: ab, covered
#sensitivity overlap
# 1) dark: ab, not targ
# 2) duds: ab, targ, not sensitive
# 3) prioritized: ab, targ, sensitive
# 4) mysterious: not ab, targ, sensitive
# 5) uninteresting: not sensitive, targ, not aberrational
}
#'@title The central function used in an overlap analysis
#'@description Examines the actual overlap between functionally significant and aberrationally signficant pathways.
#'@param functionalEnrichmentAnalysis The functional enrichment analysis to be used.
#'@param combinedAberrations The aberration analysis to be used
#'@param ola The overlap analysis runner object to be used.
#'@param coverage If coverage is limited, this will contain the set of genes the coverage is limited to.
#'@return An overlap analysis
#'@import VennDiagram
coreOverlapAnalysis<-function(functionalEnrichmentAnalysis, combinedAberrations, ola, coverage=NULL){
# save(functionalEnrichmentAnalysis, file="./output/overlapfunctional.rda")
# save(ola, file="./output/overlapOLAobject.rda")
# save(combinedAberrations, file="./output/overlapcombinedAberrations.rda")
# save(coverage, file="./output/overlapcoverage.rda")
if(is.null(coverage)) coverage=ola$results$functional_drug_screen_summary$coverage_summary
# if( is.null(functionalEnrichmentAnalysis$pathsummary) & !is.null(coverage) ){
# functionalEnrichmentAnalysis = coverage
# }
dir.create(path=ola$outfname_path, showWarnings=F, recursive=T)
outlist = list()
outlist[["imageSlots"]] = list()
drugcoverage = coverage$pathsummary[,"path_id"]
if( nullOrEmpty( functionalEnrichmentAnalysis$pathsummary ) ){
senspaths = NULL
sensen=NULL
print("There are no sensitive targets in this sample.")
cohortOrcoverage = TRUE
print(cohortOrcoverage)
}else{
print("Sensitive targets found.")
senspaths = functionalEnrichmentAnalysis$pathsummary[,"path_id"]
sensen = functionalEnrichmentAnalysis$pathsummary[functionalEnrichmentAnalysis$pathsummary[,ola$enrich_col]<ola$threshold,"path_id"]
cohortOrcoverage = nrow(functionalEnrichmentAnalysis$patientsums)>1
print(cohortOrcoverage)
}
# if(is.null(senspaths)){
# print("There are no sensitive targets in this sample.")
# sensen=NULL
# }else{
# #T3 paths enriched for drug-sensitive genes
# print("Sensitive targets found.")
# sensen = functionalEnrichmentAnalysis$pathsummary[functionalEnrichmentAnalysis$pathsummary[,ola$enrich_col]<ola$threshold,"path_id"]
# }
#vennList[["Paths enriched for drug-sensitve targets"]]
#T4 aberration_coverage
abpathsum = combinedAberrations$pathsummary
if(VERBOSE) print(names(combinedAberrations))
if(VERBOSE) print(is.null(abpathsum))
aben = abpathsum[abpathsum[,ola$enrich_col]<ola$threshold,"path_id"] #aberration enriched
if(VERBOSE) print("Creating venn diagram..")
# fullvennFileName = paste(ola$outfname_path,vennFileName,sep="")
# fullvennFileName = vennFileName
vennList = list()
vennFileName= ifelse(test=cohortOrcoverage,
no=paste0(rownames(functionalEnrichmentAnalysis$patientsums),"overlap_venn_diagram.png"),
yes="Cohort_overlap_venn_diagram.png")
vennList[["Drug-screen targeted paths"]] = drugcoverage
vennList[["Aberration-enriched\npaths"]] = aben
# print("search():")
# print(search())
# print("ls()")
# print(ls())
require(VennDiagram)
if( is.null(functionalEnrichmentAnalysis) ){#ola$results$functional_drug_screen_summary$pathsummary)){#if this is null, then there is only a drug coverage analysis available
print("venn option 1 -- functionalEnrichmentAnalysis is NULL")
vennTmp = venn.diagram(vennList,
filename=NULL,
main.cex=1.5,
cex=1.5,
cat.cex=c(1.3,1.3),
col="transparent",
cat.pos=c(340,20),
cat.dist=c(.001,.001),
cat.col=c("red","blue"),
fill=c("red","blue"),
main="Overlap of significantly aberrational and drug-targeted paths")
}else{
print("venn option 2...")
vennList[["Paths containing\ndrug-sensitive targets"]] = senspaths
vennTmp = venn.diagram(vennList, filename=NULL,
main.cex=1.5,
cex=1.5,
cat.cex=c(1.3,1.3,1.3)[1:length(vennList)],
col="transparent",
cat.pos=c(340,30,0)[1:length(vennList)],
cat.dist=c(.001,.001,-.001)[1:length(vennList)],
cat.col=c("red","darkorchid4","blue")[1:length(vennList)],
fill=c("red","darkorchid4","blue")[1:length(vennList)],
main="Overlap of significantly aberrational,\ndrug-targeted and drug sensitive paths")
print("venn built..")
}
plot.new()
print("..")
grid.draw(vennTmp)
print("saving plot...")
print(vennFileName)
fullvennFileName = save.plot(pname=vennFileName)
# png(fullvennFileName)
# dev.off()
print(fullvennFileName)
if(VERBOSE) print(vennFileName)
outlist$imageSlots[[vennFileName]] = fullvennFileName
print("Venn diagram complete.")
# #this is the base name (no folder root) of the venn diagram;
# basename = strsplit(vennFileName,split="\\/")[[1]][length(strsplit(vennFileName,split="\\/")[[1]])]
# outlist[[basename]] = basename #setting up a base name with a .png should compel toHTML to insert an image. .. lets see if it works
#
if(VERBOSE) print("Creating overlap tables...")
#overlap tables
######1
abNotDrug = setdiff(aben, drugcoverage)#aberration enriched, not drug targeted
if(VERBOSE) print("tabout_abNotDrug")
if(VERBOSE) print(length(abNotDrug))
if(VERBOSE) print(dim(combinedAberrations$pathsummary))
if(VERBOSE) print(is.null(combinedAberrations$pathsummary))
tabout_abNotDrug = combinedAberrations$pathsummary[abNotDrug,,drop=F]
if(VERBOSE) print("got combinedAberrations$pathsummary[abNotDrug,,drop=F]")
outlist[["Aberration enriched, not drug targeted"]] = tabout_abNotDrug
if(VERBOSE) print("set tabout_abNotDrug")
if(length(tabout_abNotDrug)>0){
outlist[["Pathway overlaps of genes in aberration enriched, not drug targeted paths"]]=BangForBuck(path_detail=ola$path_detail,darkPaths=tabout_abNotDrug)
}else{
outlist[["Pathway overlaps of genes in aberration enriched, not drug targeted paths"]]="No dark pathways were found (ie, no pathways were found to be aberrationally enriched but not drug targeted)"
}
if( VERBOSE ){
print("targAndAb")
######2
print(is.null(aben))
print(is.null(drugcoverage))
}
targAndAb = intersect(drugcoverage,aben)#Aberrationally enriched, containing drug targets
if(VERBOSE) print("tabout_targAndAb")
if(VERBOSE) print(is.null(targAndAb))
tabout_targAndAb = NULL
if( !is.null(targAndAb) ){
tabout_targAndAb = merge(coverage$pathsummary[targAndAb,],
combinedAberrations$pathsummary[targAndAb,],
by="path_id",
all=F,
sort=F)
if(VERBOSE) print("Checking targeted and aberrational overlap..")
if(nrow(tabout_targAndAb)){
if(VERBOSE) print("Targeted and aberrational found..")
cohortOrPatient="cohort"
if( !is.null(functionalEnrichmentAnalysis$patientsums) ){
if( nrow(functionalEnrichmentAnalysis$patientsums)==1 ){
cohortOrPatient = rownames(functionalEnrichmentAnalysis$patientsums)
}
}
plotfname = paste("patient",cohortOrPatient,"aberrational_and_targeted.png")
# fullplotfname = paste(ola$outfname_path,plotfname, sep="")
#show a bubble plot of the pathways that are targeted and aberrational
tmpPlotFName = plotOverlapArea(tabout_targAndAb=tabout_targAndAb,
results=ola$results,
abandtargplotname=plotfname)
outlist$imageSlots[[plotfname]] = tmpPlotFName
tabout_targAndAb=cbind(1:nrow(tabout_targAndAb),tabout_targAndAb)
colnames(tabout_targAndAb)[1]<-"path number"
if(VERBOSE) print("Plot created.")
}
}else{
readline("not going to plot the overlap..")
}
outlist[["Aberrationally enriched, containing drug targets"]] = tabout_targAndAb
if(VERBOSE) print("drugNotAb")
######3
drugNotAb = setdiff(drugcoverage, aben)#drug targeted, not aberrationally enriched
if(VERBOSE) print("drugNotAb; merging to make tabout_drugNotAb")
if(VERBOSE) print(dim(coverage$pathsummary[drugNotAb,]))
if(VERBOSE) print(dim(combinedAberrations$pathsummary))
if( is.null(dim(combinedAberrations$pathsummary)) ){
if(VERBOSE) print("it's null")
if(VERBOSE) print(combinedAberrations$pathsummary)
tabout_drugNotAb = "Patient was not found to have aberrations in currently employed set of pathways."
if(VERBOSE) print("Patient was not found to have aberrations in currently employed set of pathways.")
}else{
tabout_drugNotAb = merge(x=coverage$pathsummary[drugNotAb,],
y=combinedAberrations$pathsummary,
by="path_id",
all.x=T,
sort=F)#merges so that all rows with drug screen summary will be included
if(VERBOSE) print(names(combinedAberrations))
if(VERBOSE) print("merged; zeroing the na's")
tabout_drugNotAb[is.na(tabout_drugNotAb[,11]),11]<- 0
hyperg_p_value_indexes = grep(pattern="hyperg_p_value",x=colnames(tabout_drugNotAb),ignore.case=T)
hyperg_p_value_index = hyperg_p_value_indexes[length(hyperg_p_value_indexes)]
#sort targeted, not aberrationally enriched by the proportion aberrational
tabout_drugNotAb = tabout_drugNotAb[order(tabout_drugNotAb[,hyperg_p_value_index],decreasing=F),]
}
outlist[["Drug targeted, not aberrationally enriched"]] = tabout_drugNotAb
allSensPathsMergAb = "No pathways were found to contain drug-sensitive genes"
if( !nullOrEmpty(functionalEnrichmentAnalysis$pathsummary) ){
allSensPathsMergAb = functionalEnrichmentAnalysis$pathsummary
if(VERBOSE) print("allSensPathsMergAb; merging to make allSensPathsMergAb")
if (is.null(dim(combinedAberrations$pathsummary)) ){
print("Patient was not found to have aberrations in currently employed set of pathways.")
}else{
tabout_allSensPathsMergAb = merge(x=functionalEnrichmentAnalysis$pathsummary,
y=combinedAberrations$pathsummary,
by="path_id",
all.x=T,
sort=F)#merges so that all rows with drug screen summary will be included
if(VERBOSE) print("merged; zeroing the na's")
tabout_allSensPathsMergAb[is.na(tabout_allSensPathsMergAb[,11]),11]<- 0
hyperg_p_value_indexes = grep(pattern="hyperg_p_value",x=colnames(tabout_allSensPathsMergAb),ignore.case=T)
hyperg_p_value_index = hyperg_p_value_indexes[length(hyperg_p_value_indexes)]
#sort targeted, not aberrationally enriched by the proportion aberrational
tabout_allSensPathsMergAb = tabout_allSensPathsMergAb[order(tabout_allSensPathsMergAb[,hyperg_p_value_index],decreasing=F),]
allSensPathsMergAb = tabout_allSensPathsMergAb
}
}
outlist[["Paths containing drug-sensitive genes"]] = allSensPathsMergAb
if(VERBOSE) print("sensenAndAb")
if( is.null(aben) ){
outlist[["Enriched for aberration and enriched for sensitive drug targets"]] = "Patient was not found to have aberrations in currently employed set of pathways."
outlist[["Aberration enriched, containing sensitive targets"]] = "Patient was not found to have aberrations in currently employed set of pathways."
}else{
sensenAndAb = intersect(sensen,aben)#enriched for sensitive targets and enriched for aberration
outlist[["Enriched for aberration and enriched for sensitive drug targets"]] = as.matrix(sensenAndAb,ncol=1)
if(VERBOSE) print("conSensAndAb")
conSensAndAb = intersect(senspaths,aben)#paths containing sensitive genes, that are also aberration enriched
outlist[["Aberration enriched, containing sensitive targets"]] = as.matrix(conSensAndAb, ncol=1)
}
# cat("\nWriting set of overlap-summary text files to:", paste("overlap_summary", as.character(Sys.time())), "\n")
# outHTMLname = paste(outfname, ".html",sep="")
#cat("\nWriting html-based summary to ", outHTMLname,"\n")
ret = list()
outlist[["combined_aberrations_summary"]] = combinedAberrations
outlist[["functional_drug_screen_summary"]] = functionalEnrichmentAnalysis
if(VERBOSE){
print(names(combinedAberrations))
print(names(combinedAberrations$path_summary_each_patient))
}
return(outlist)
}#coreOverlapAnalysis
#nullOrEmpty
#checks if an object, dat, is either null or empty, with zero rows or zero length
nullOrEmpty<-function(dat){
if(is.null(dat)) return(TRUE)
if(class(dat)=="matrix"|class(dat)=="data.frame"){
if(!nrow(dat)) return(TRUE)
}else{
if(!length(dat)) return(TRUE)
}
return(FALSE)
}
# olatest = abDrugOverlapAnalysis(results=results, path_detail=path_detail)
biodev/packageDir documentation built on Nov. 4, 2019, 7:19 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions RunOverlapAnalysis test.abDrugOverlapAnalysis abDrugOverlapAnalysis selectOverlapType runUnmatchedCohort pickSettings runMatchedCohort runMatchedIndividuals runUnmatchedIndividuals CheckPatientOverlap plotOverlapArea adjustOverlapToSensitive extractedDrugScreenAnalysis combineAbSens coreOverlapAnalysis2 coreOverlapAnalysis nullOrEmpty
Documented in abDrugOverlapAnalysis coreOverlapAnalysis
#overlapAnalaysisFunctions
RunOverlapAnalysis<-function(settings, study){
results = abDrugOverlapAnalysis(study=study,
settings=settings)
return(results)
}
test.abDrugOverlapAnalysis<-function(){
#stud = STUDY
#test all instances with individuals and cohorts.
#test matched cohort
#get default overlap analysis settings
defSettings = getTestStudySettings()
overlapSettings = defSettings$overlap_analysis
#get a study object pre-loaded with data aberration and functional data.
stud = getTestStudyObject()
#run the overlap analysis with functional genomic and genomic aberration data from the same cohort
overlapRes1 = abDrugOverlapAnalysis(study=stud, settings=overlapSettings)
#simulate the analysis with functional genomic and genomic aberration data from sepparate cohorts
studySepparateCohort = stud
patientSums = studySepparateCohort@results$functional_drug_screen_summary$patientsums
rownames(patientSums) <- toupper(rownames(patientSums))
studySepparateCohort@results$functional_drug_screen_summary$patientsums = patientSums
overlapRes2 = abDrugOverlapAnalysis(study=studySepparateCohort,
settings=overlapSettings)
}
#settings slots expected:
# combinedAberrations
#'@title abDrugOverlapAnalysis
#'@description main function for running analysis of functional and genomic aberration arms
#'@param study A study object (must have results loaded for aberration and functional analyses)
#'@param thresh Numeric, the threshold used to establish which pathways will be considered to be drug sensitive and aberrational.
#'@param enrich_col String, the name of the column in the path summary used to determine significance of a pathway. This is the column to which the threshold argument is to be applied.
#'@param verbose Logical, a flag indicating if the program is to provide extra diagnostic output.
#'@param settings List obejct. This contains the settings used in the overlap analysis. Use getDefaultOverlapSettings()
#'@return Study object with overlap analysis in results list added or replaced
#'@export
#'@examples
#'#get a study object pre-loaded with genomic aberration and functional genomic data.
#' stud = getTestStudyObject()
#'
#'#get default overlap analysis settings
#'defSettings = getTestStudySettings()
#'overlapSettings = defSettings$overlap_analysis
#'
#'#run the overlap analysis
#'
#'overlapRes1 = abDrugOverlapAnalysis(study=stud, settings=overlapSettings)
#'
#'\dontrun{
#'#If settings for the overlap analysis cannot be found, the overlap
#'#analysis will run in interactive mode, requiring user input.
#'#This line of code will run the overlap anaysis in interactive mode,
#'#requesting user input to establish settings:
#'
#'abDrugOverlapAnalysis(study=stud)
#'}
#'@importFrom VennDiagram venn.diagram
#'@importFrom calibrate textxy
abDrugOverlapAnalysis<-function(study,
thresh=0.05,
enrich_col="hyperg_p_w_FDR",
verbose=T,
settings=NULL){
if(!exists("VERBOSE")) VERBOSE=T
cat("\n******************************************************************************************\n")
cat("\n*************************************overlap analysis*************************************\n")
cat("\n******************************************************************************************\n")
checkPathsMatch(STUDY=study)
sfolder = paste(studyFolder(s=study),"/results/", sep="")
#set up ola object
ola = list() #ola = overlap analysis
ola$results = study@results
ola$path_detail = getPaths(study)
ola$threshold = thresh
ola$enrich_col = enrich_col
ola$outfname_path = sfolder
ola$verbose = verbose
if("interactive"%in%names(settings)){
#if interactive is found as a flag in settings it is the first time overlap analysis has been run for the study
#put it in the right place
if( "basic_overlap"%in%names(settings) ){#overlap analysis has already been run
settings_full = settings #transfer the settings list
settings_full$interactive = NULL #erase the flag from the master list
interactiveFlag = settings$interactive # put the interactive flag in the right place
settings = settings_full$basic_overlap # in the
settings$interactive = interactiveFlag
}else{#it's the first time the overlap was run on the current study
settings_full = list()
settings_full$basic_overlap = settings
}
}else{
settings_full = settings
settings = settings_full$basic_overlap
}
# settings$interactive = settings_full$interactive
# settings_full$interactive = NULL
settings = setting(s=settings, prompt="Would you like to run the function/aberration overlap analysis on each patient? (y/n)")
runEachPatient=settings$.text == "y"
ola$runEachPatient = runEachPatient
while(T){
settings = setting(s=settings, "Would you like to use default path analysis settings or use those associated with study arms? (please enter d or a)")
if(settings$.text%in%c("a","d")) break
print("Sorry, your input was not understood, please try again.")
}
if( settings$.text=="d" ){
settings_full$defaultSummaryTable = getDefaultSettings()$defaultSummaryTable
settings$interactive = F
}else{
settings_full$defaultSummaryTable = NULL
settings$interactive = T
}
settings_full$basic_overlap = settings
ola$settings = settings_full
ola = selectOverlapType(ola=ola, study=study)
#system('/usr/bin/afplay ./reference_data/Submarine.aiff')
ola$results$overlap_analysis$settings = ola$settings
return(ola$results$overlap_analysis)
}
selectOverlapType<-function(ola, study){
#determine if matched
overlapPatients = CheckPatientOverlap(results=ola$results)
#Assure only the sensitive patients are
overlapPatients = adjustOverlapToSensitive(overlap=overlapPatients,
functionalEnrichmentAnalysis=ola$results$functional_drug_screen_summary)
matched = length(overlapPatients)>0 #if there are overlap patients, it's assumed to be a matched sample
if( matched ){
#1: cohort, matched
#limit cohort
ola = runMatchedCohort(overlapPatients=overlapPatients, ola=ola, study=study)
}else{
cat("\nNote: no patient sample ids were found to be associated with\nboth sensitive gene targets and aberration analysis/analyses.\n")
#2: cohort, unmatched
#no limiting, just run it once the combination is done
ola = runUnmatchedCohort(ola=ola, study=study)
}
if( ola$runEachPatient ){#if the overlap analysis should be run for each patient
ola$outfname_path = paste(ola$outfname_path, "overlap_analysis_each_patient/", sep="")
if(matched){
#3: individual patient, matched
#pass drug screen summary for patient plus aberration summary for patient
ola = runMatchedIndividuals(overlapPatients=overlapPatients, ola=ola, study=study)
}else{
#4: individual patient, unmatched
#pass drug screen summary for individual patient, plus aberration summary for whole cohort
ola = runUnmatchedIndividuals(ola=ola, study=study)
}
#indresults should be the list with each slot named with a patient identifier
}
return(ola)
}
runUnmatchedCohort<-function(ola, study, functional_analysis_name="functional_drug_screen_summary"){
if(VERBOSE) cat("\nola$settings:\n")
if(VERBOSE) print(ola$settings)
functionalEnrichmentAnalysis = ola$results[[functional_analysis_name[1]]]
#case 1: use what was passed as settings
# combSettings = ola$settings$combinedAberrations
if(is.null(ola$settings$defaultSummaryTable)){#if the default settings aren't provided, look for the settings in the arms
ola$settings$combinedAberrations = NULL
#case 2: use the default settings
#case 3: use the settings from an aberration arm (taken care of in combineAberrationTypes)
# ---> if the user said not to use the default, the combineAberrationTypes2 will make the user pick
}else{
ola$settings$combinedAberrations = ola$settings$defaultSummaryTable
}
if(is.null(ola$results$combinedAberrations)) ola$results$combinedAberrations = combineAberrationTypes(study=study,
s=ola$settings$combinedAberrations,
results=ola$results,
runEachPatient=F)
ola$settings$combinedAberrations = ola$results$combinedAberrations$settings
combinedAberrations = ola$results$combinedAberrations
ola$outfname_path = paste(ola$outfname_path, "overlap_analysis/", sep="")
ola$results$overlap_analysis = coreOverlapAnalysis(functionalEnrichmentAnalysis=functionalEnrichmentAnalysis,
combinedAberrations=combinedAberrations,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
return(ola)
}#runUnmatchedCohort
pickSettings<-function(study, settingType = "aberration"){
si = grep(pattern=settingType, x=names(study@studyMetaData@settings), ignore.case=T)
targetArmSettings = names(study@studyMetaData@settings)[si]
if(length(si)>1){
sres = settingList(s=list(interactive=T),
prompt="Please pick a study arm whose path analysis settings you would like to use.",
set=names(study@studyMetaData@settings)[si])
targetArmSettings = sres$.text
}
return(study@studyMetaData@settings[[targetArmSettings]])
}
runMatchedCohort<-function(ola, overlapPatients, study){
# if(is.null(ola$settings$combinedAberrations)) ola$settings$combinedAberrations =
#case 1: use the settings passed
funSettings = ola$settings$functionalPathAnalysis
if(is.null(ola$settings$defaultSummaryTable)){#if case 1 is null
ola$settings$functionalPathAnalysis = NULL
#case 2: use the default settings
#case 3: use the settings from an aberration arm (taken care of in combineAberrationTypes)
# ---> if the user said not to use the default, the combineAberrationTypes2 will make the user pick
}else{
ola$settings$functionalPathAnalysis = ola$settings$defaultSummaryTable
}
cat("\n\nAnalysis of functional drug screen for patients who have had both aberration and functional analysis..\n")
functionalEnrichmentAnalysis = extractedDrugScreenAnalysis(psubset=overlapPatients,
study=study,
s=ola$settings$functionalPathAnalysis,
results=ola$results,
path_detail=ola$path_detail,
eachPatient=ola$runEachPatient) #holds the results for the functional enrichment analysis
ola$results$overlap_analysis$functional_drug_screen_summary = functionalEnrichmentAnalysis
ola$settings$functionalPathAnalysis = functionalEnrichmentAnalysis$settings
# overlapPatients = intersect(overlapPatients, colnames(functionalEnrichmentAnalysis$patientsums))
if(is.null(ola$results$combinedAberrations)){
#case 1: use the settings passed
if(is.null(ola$settings$defaultSummaryTable)){#if the default settings aren't provided, look for the settings in the arms
ola$settings$combinedAberrations = NULL
#case 2: use the default settings
#case 3: use the settings from an aberration arm (taken care of in combineAberrationTypes)
# ---> if the user said not to use the default, the combineAberrationTypes2 will make the user pick
}else{
ola$settings$combinedAberrations = ola$settings$defaultSummaryTable
}
cat("\n\nAnalysis of combined aberration data for patients who have had both aberration and functional analysis..\n")
ola$results$combinedAberrations = combineAberrationTypes2(study=study,
s=ola$settings$combinedAberrations,
results=ola$results,
overlapPatients=overlapPatients)
}
ola$settings$combinedAberrations = ola$results$combinedAberrations$settings
ola$outfname_path = paste(ola$outfname_path, "overlap_analysis/", sep="")
ola$results$overlap_analysis = coreOverlapAnalysis(functionalEnrichmentAnalysis=functionalEnrichmentAnalysis,
combinedAberrations=ola$results$combinedAberrations,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
return(ola)
}
runMatchedIndividuals<-function(ola, overlapPatients, study){
# functionalEnrichmentAnalysis = NULL #holds the results for the functional enrichment analysis
# combinedAberrations = NULL #holds the results for the combined aberration enrichment analysis
pol = list() #patient overlap analyses
tmpAbs = ola$settings$combinedAberrations
#pull out the appropriate summaries
if(is.null(tmpAbs)){
tmpAbs = ola$results$combinedAberrations$settings
ola$settings$combinedAberrations = tmpAbs
}
tmpAbs$"To limit by count of patients with genes in examined state, enter count here:\n(Enter 1 to skip limiting by count)" = "1"
tmpAbs$'To limit by proportion of cohort with genes in examined state, enter minimum proportion here:\n(enter 0 for no limit): '="0"
tmpDs = ola$settings$functional_drug_screen_summary
if(is.null(tmpDs)){
tmpDs = ola$settings$defaultSummaryTable
# tmpDs = ola$results$functional_drug_screen_summary$settings
# if(class(tmpDs)!="list") tmpDs = dfToList(df=tmpDs)
# tmpDs$interactive = ola$settings$basic_overlap$interactive
ola$settings$functional_drug_screen_summary = tmpDs
}
tmpDs$"To limit by count of patients with genes in examined state, enter count here:\n(Enter 1 to skip limiting by count)" = "1"
tmpDs$'To limit by proportion of cohort with genes in examined state, enter minimum proportion here:\n(enter 0 for no limit): '="0"
fpathtmp = ola$outfname_path
for(p in overlapPatients){
print(p)
ola$outfname_path=paste(fpathtmp, p,"/overlap_analysis/", sep="")
combsum = combineAberrationTypes2(results=ola$results,
study=study,
s=tmpAbs,
overlapPatients=p)
functionalSum = extractedDrugScreenAnalysis(psubset=p,
eachPatient=F,
study=study,
s=tmpDs,
results=ola$results,
path_detail=ola$path_detail)
#run the overlap analysis
pres = coreOverlapAnalysis(functionalEnrichmentAnalysis=functionalSum,
combinedAberrations=combsum,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
#put the results in the right place
pol[[p]] = list()
pol[[p]][["overlap_analysis"]] = pres
}
ola$results$overlap_analysis$overlap_analysis_each_patient = pol
return(ola)
}#runMatchedIndividuals
runUnmatchedIndividuals<-function(ola, study){
functionalEnrichmentAnalysis = ola$results$overlap_analysis$functional_drug_screen_summary #holds the results for the functional enrichment analysis
combinedAberrations = ola$results$overlap_analysis$combined_aberrations_summary #holds the results for the combined aberration enrichment analysis
#get the set of individuals to be analyzed
patientSet = rownames(ola$results$functional_drug_screen_summary$patientsums)
ola$results$overlap_analysis$overlap_analysis_each_patient = list()
pol=list()
fpathtmp = ola$outfname_path
for(p in patientSet){
print(p)
ola$outfname_path=paste(fpathtmp, p,"/overlap_analysis/", sep="")
#extract patient record, extract patient drug results
patientFunctionalAnalysis = functionalEnrichmentAnalysis$path_summary_each_patient[[p]]
pres = coreOverlapAnalysis(functionalEnrichmentAnalysis=patientFunctionalAnalysis,
combinedAberrations=combinedAberrations,
ola=ola,
coverage=ola$results$functional_drug_screen_summary$coverage_summary)
pol[[p]] = list()
pol[[p]][["overlap_analysis"]] = pres
}
ola$results$overlap_analysis$overlap_analysis_each_patient = pol
return(ola)
}#runUnmatchedIndividuals
CheckPatientOverlap<-function(results){
cat("..checking patients with functional and aberrational analysis..")
psetOut = c()
#check if analysis will be limited to patients who are in both cohorts, the drugs screen and the aberration sets
funIndex = grep("functional", x=names(results))
psetOut = rownames(results[[funIndex]]$patientsums)
abIndex = grep("aberration_", x=names(results))
for(i in abIndex){
psetOut = intersect(x=psetOut, y=rownames(results[[i]]$patientsums))
}
cat(" ..checked...")
return(psetOut)
}#CheckPatientOverlap
# load("./testsSavePlotOverlapData.rda", verbose=T)
# tabout_targAndAb = allInputs$tabout_targAndAb
# results = allInputs$results
# abandtargplotname = allInputs$abandtargplotname
plotOverlapArea<-function(tabout_targAndAb, results, abandtargplotname, autosave=T){
# readline("inside plot overlap area()")
# allInputs = list(tabout_targAndAb=tabout_targAndAb, results=results, abandtargplotname=abandtargplotname)
# save(allInputs, file="./testsSavePlotOverlapData.rda")
if(!nrow(tabout_targAndAb)){
print("There are no pathways in the overlap between drug targeted/drug sensitive and aberrational")
return()
}
#######################
#### plot overlap area
if(VERBOSE) print("Plotting overlap area.. ")
if(VERBOSE) print(colnames(tabout_targAndAb))
# tabout_targAndAb$path_id = gsub(pattern=" ",replacement="\n",x=tabout_targAndAb$path_id)
p1=ggplot(data=tabout_targAndAb, aes(x=proportion_of_cohort_w_aberrational_gene_in_path,
y=drug_targeted_genes,
size=testable_path_length.x,
label=path_id))+
geom_point(alpha=0.5)+
scale_size(range=c(5,30), guide = guide_legend(title = "Number of genes in path"))+
geom_text(size=2, vjust=1)+
xlab("Proportion of cohort with aberrational gene in path")+
ylab("Number of drug targeted genes")+
ggtitle("Proportion affected vs number of drug targets")+
theme_bw()
print(p1)
if(autosave){
pfname = plotFileName(abandtargplotname)
ggsave(plot=p1, width=10, height=6, filename=pfname)
}
# png(filename=abandtargplotname,width=800,height=500)
# maxpathlength = max(tabout_targAndAb[,3])
# cexvals = (tabout_targAndAb[,3]/maxpathlength)*10
#
# plot(cex=cexvals,
# ylim=c(0,1.2*max(tabout_targAndAb[,"drug_targeted_genes"])),#)),
# x=tabout_targAndAb[,"proportion_of_cohort_w_aberrational_gene_in_path"],
# y=tabout_targAndAb[,"drug_targeted_genes"],#"drug_targeted_genes"],
# main="Proportion affected vs number of drug targets.",
# xlab="Proportion of cohort with aberrational gene in path",
# ylab="Number of drug targets")
# textxy(tabout_targAndAb[,"proportion_of_cohort_w_aberrational_gene_in_path"],
# tabout_targAndAb[,"drug_targeted_genes"],
# labs=1:nrow(tabout_targAndAb))#,
# #cx=1.3)
# legend(x="topright",
# legend=paste("Size = path length (",paste(range(tabout_targAndAb[,3]), collapse=" to "),"nodes)\nNumber = path number in the \'enriched and targeted\' table"),
# cex=.8, pch=1)
# dev.off()
return(pfname)
}#plotOverlapArea
adjustOverlapToSensitive<-function(overlap, functionalEnrichmentAnalysis){
#get the total sensitive genes per patient
psums = functionalEnrichmentAnalysis$patientsums
#extract the patient ids for those patients with active genes
keepers = rownames(psums)[psums[,1]>0]
#subset the overlap patients to only those patients with sensitive genes
newOverlap = intersect(overlap, keepers)
return(newOverlap)
}#adjustOverlapToSensitive
extractedDrugScreenAnalysis<-function(psubset, results, path_detail, study, s, eachPatient){
#extractedDrugScreenAnalysis
#runs path analysis on drug screen patient subset
#returns drug screen results set with coverage analysis appended
#first get the functional analysis out of results
cat("...inside extractedDrugScreenAnalysis()...")
findex=grep(pattern="functional", names(results),ignore.case=T)
if(length(findex)>1) warning("More than one functional analysis was found in the input to the overlap analysis. (ex: more than one drug screen) This is currently a functionality of the program which has not been tested!")
if(VERBOSE) print(names(results))
if(VERBOSE) print(findex)
fa = results[[findex]]
pgm = as.matrix(fa$patientGeneMatrix[,psubset,drop=F]) #extract the subset of patients from the patient gene matrix
covset = rownames(fa$coverage_summary$genesummary)
# tm = getTargetMatrix(tgenes=covset, paths=path_detail$paths)
if(VERBOSE) print("summaryTable from extractedDrugScreenAnalysis")
cat("\ndim pgm going into summaryTable:", dim(pgm), "\n")
if(!is.null(pgm) & (nrow(pgm)>0)){
if(is.null(s)) s = fa$settings
summ = summaryTable(study=study,
coverageGeneDescription="targeted",
coverageDataSetDescription="Coverage of functional analysis platform",
individualEnrichment=eachPatient,
coverage=covset,
settings=s,
pgm=pgm,
activeGeneDescription="functionally_affected",
dataSetDescription="Functional path enrichment in patient subset for function-aberration overlap analysis")
}else{
print("There were no sensitive genes in this patient")
summ = list()
}
summ$coverage_summary = fa$coverage_summary
return(summ)
}#extractedDrugScreenAnalysis
combineAbSens<-function(functionalEnrichmentAnalysis, combinedAberrations, ola, coverage=NULL){
}
coreOverlapAnalysis2<-function(functionalEnrichmentAnalysis, combinedAberrations, ola, coverage=NULL){
### combine into single table:
# functional path enrichments
# aberration path enrichments
# functional coverage
# aberration data coverage
combtab = combineAbSens(functionalEnrichmentAnalysis=functionalEnrichmentAnalysis, combinedAberrations=combinedAberrations, ola=ola, coverage=coverage)
### determine what kind of overlap
# 1) ab + sensitivity + panel coverage
# 2) ab + panel coverage
### find which paths are in each part of the overlap
#panel coverage:
# 1) dark: ab, not covered
# 2) light: covered, not ab
# 3) prioritized: ab, covered
#sensitivity overlap
# 1) dark: ab, not targ
# 2) duds: ab, targ, not sensitive
# 3) prioritized: ab, targ, sensitive
# 4) mysterious: not ab, targ, sensitive
# 5) uninteresting: not sensitive, targ, not aberrational
}
#'@title The central function used in an overlap analysis
#'@description Examines the actual overlap between functionally significant and aberrationally signficant pathways.
#'@param functionalEnrichmentAnalysis The functional enrichment analysis to be used.
#'@param combinedAberrations The aberration analysis to be used
#'@param ola The overlap analysis runner object to be used.
#'@param coverage If coverage is limited, this will contain the set of genes the coverage is limited to.
#'@return An overlap analysis
#'@import VennDiagram
coreOverlapAnalysis<-function(functionalEnrichmentAnalysis, combinedAberrations, ola, coverage=NULL){
# save(functionalEnrichmentAnalysis, file="./output/overlapfunctional.rda")
# save(ola, file="./output/overlapOLAobject.rda")
# save(combinedAberrations, file="./output/overlapcombinedAberrations.rda")
# save(coverage, file="./output/overlapcoverage.rda")
if(is.null(coverage)) coverage=ola$results$functional_drug_screen_summary$coverage_summary
# if( is.null(functionalEnrichmentAnalysis$pathsummary) & !is.null(coverage) ){
# functionalEnrichmentAnalysis = coverage
# }
dir.create(path=ola$outfname_path, showWarnings=F, recursive=T)
outlist = list()
outlist[["imageSlots"]] = list()
drugcoverage = coverage$pathsummary[,"path_id"]
if( nullOrEmpty( functionalEnrichmentAnalysis$pathsummary ) ){
senspaths = NULL
sensen=NULL
print("There are no sensitive targets in this sample.")
cohortOrcoverage = TRUE
print(cohortOrcoverage)
}else{
print("Sensitive targets found.")
senspaths = functionalEnrichmentAnalysis$pathsummary[,"path_id"]
sensen = functionalEnrichmentAnalysis$pathsummary[functionalEnrichmentAnalysis$pathsummary[,ola$enrich_col]<ola$threshold,"path_id"]
cohortOrcoverage = nrow(functionalEnrichmentAnalysis$patientsums)>1
print(cohortOrcoverage)
}
# if(is.null(senspaths)){
# print("There are no sensitive targets in this sample.")
# sensen=NULL
# }else{
# #T3 paths enriched for drug-sensitive genes
# print("Sensitive targets found.")
# sensen = functionalEnrichmentAnalysis$pathsummary[functionalEnrichmentAnalysis$pathsummary[,ola$enrich_col]<ola$threshold,"path_id"]
# }
#vennList[["Paths enriched for drug-sensitve targets"]]
#T4 aberration_coverage
abpathsum = combinedAberrations$pathsummary
if(VERBOSE) print(names(combinedAberrations))
if(VERBOSE) print(is.null(abpathsum))
aben = abpathsum[abpathsum[,ola$enrich_col]<ola$threshold,"path_id"] #aberration enriched
if(VERBOSE) print("Creating venn diagram..")
# fullvennFileName = paste(ola$outfname_path,vennFileName,sep="")
# fullvennFileName = vennFileName
vennList = list()
vennFileName= ifelse(test=cohortOrcoverage,
no=paste0(rownames(functionalEnrichmentAnalysis$patientsums),"overlap_venn_diagram.png"),
yes="Cohort_overlap_venn_diagram.png")
vennList[["Drug-screen targeted paths"]] = drugcoverage
vennList[["Aberration-enriched\npaths"]] = aben
# print("search():")
# print(search())
# print("ls()")
# print(ls())
require(VennDiagram)
if( is.null(functionalEnrichmentAnalysis) ){#ola$results$functional_drug_screen_summary$pathsummary)){#if this is null, then there is only a drug coverage analysis available
print("venn option 1 -- functionalEnrichmentAnalysis is NULL")
vennTmp = venn.diagram(vennList,
filename=NULL,
main.cex=1.5,
cex=1.5,
cat.cex=c(1.3,1.3),
col="transparent",
cat.pos=c(340,20),
cat.dist=c(.001,.001),
cat.col=c("red","blue"),
fill=c("red","blue"),
main="Overlap of significantly aberrational and drug-targeted paths")
}else{
print("venn option 2...")
vennList[["Paths containing\ndrug-sensitive targets"]] = senspaths
vennTmp = venn.diagram(vennList, filename=NULL,
main.cex=1.5,
cex=1.5,
cat.cex=c(1.3,1.3,1.3)[1:length(vennList)],
col="transparent",
cat.pos=c(340,30,0)[1:length(vennList)],
cat.dist=c(.001,.001,-.001)[1:length(vennList)],
cat.col=c("red","darkorchid4","blue")[1:length(vennList)],
fill=c("red","darkorchid4","blue")[1:length(vennList)],
main="Overlap of significantly aberrational,\ndrug-targeted and drug sensitive paths")
print("venn built..")
}
plot.new()
print("..")
grid.draw(vennTmp)
print("saving plot...")
print(vennFileName)
fullvennFileName = save.plot(pname=vennFileName)
# png(fullvennFileName)
# dev.off()
print(fullvennFileName)
if(VERBOSE) print(vennFileName)
outlist$imageSlots[[vennFileName]] = fullvennFileName
print("Venn diagram complete.")
# #this is the base name (no folder root) of the venn diagram;
# basename = strsplit(vennFileName,split="\\/")[[1]][length(strsplit(vennFileName,split="\\/")[[1]])]
# outlist[[basename]] = basename #setting up a base name with a .png should compel toHTML to insert an image. .. lets see if it works
#
if(VERBOSE) print("Creating overlap tables...")
#overlap tables
######1
abNotDrug = setdiff(aben, drugcoverage)#aberration enriched, not drug targeted
if(VERBOSE) print("tabout_abNotDrug")
if(VERBOSE) print(length(abNotDrug))
if(VERBOSE) print(dim(combinedAberrations$pathsummary))
if(VERBOSE) print(is.null(combinedAberrations$pathsummary))
tabout_abNotDrug = combinedAberrations$pathsummary[abNotDrug,,drop=F]
if(VERBOSE) print("got combinedAberrations$pathsummary[abNotDrug,,drop=F]")
outlist[["Aberration enriched, not drug targeted"]] = tabout_abNotDrug
if(VERBOSE) print("set tabout_abNotDrug")
if(length(tabout_abNotDrug)>0){
outlist[["Pathway overlaps of genes in aberration enriched, not drug targeted paths"]]=BangForBuck(path_detail=ola$path_detail,darkPaths=tabout_abNotDrug)
}else{
outlist[["Pathway overlaps of genes in aberration enriched, not drug targeted paths"]]="No dark pathways were found (ie, no pathways were found to be aberrationally enriched but not drug targeted)"
}
if( VERBOSE ){
print("targAndAb")
######2
print(is.null(aben))
print(is.null(drugcoverage))
}
targAndAb = intersect(drugcoverage,aben)#Aberrationally enriched, containing drug targets
if(VERBOSE) print("tabout_targAndAb")
if(VERBOSE) print(is.null(targAndAb))
tabout_targAndAb = NULL
if( !is.null(targAndAb) ){
tabout_targAndAb = merge(coverage$pathsummary[targAndAb,],
combinedAberrations$pathsummary[targAndAb,],
by="path_id",
all=F,
sort=F)
if(VERBOSE) print("Checking targeted and aberrational overlap..")
if(nrow(tabout_targAndAb)){
if(VERBOSE) print("Targeted and aberrational found..")
cohortOrPatient="cohort"
if( !is.null(functionalEnrichmentAnalysis$patientsums) ){
if( nrow(functionalEnrichmentAnalysis$patientsums)==1 ){
cohortOrPatient = rownames(functionalEnrichmentAnalysis$patientsums)
}
}
plotfname = paste("patient",cohortOrPatient,"aberrational_and_targeted.png")
# fullplotfname = paste(ola$outfname_path,plotfname, sep="")
#show a bubble plot of the pathways that are targeted and aberrational
tmpPlotFName = plotOverlapArea(tabout_targAndAb=tabout_targAndAb,
results=ola$results,
abandtargplotname=plotfname)
outlist$imageSlots[[plotfname]] = tmpPlotFName
tabout_targAndAb=cbind(1:nrow(tabout_targAndAb),tabout_targAndAb)
colnames(tabout_targAndAb)[1]<-"path number"
if(VERBOSE) print("Plot created.")
}
}else{
readline("not going to plot the overlap..")
}
outlist[["Aberrationally enriched, containing drug targets"]] = tabout_targAndAb
if(VERBOSE) print("drugNotAb")
######3
drugNotAb = setdiff(drugcoverage, aben)#drug targeted, not aberrationally enriched
if(VERBOSE) print("drugNotAb; merging to make tabout_drugNotAb")
if(VERBOSE) print(dim(coverage$pathsummary[drugNotAb,]))
if(VERBOSE) print(dim(combinedAberrations$pathsummary))
if( is.null(dim(combinedAberrations$pathsummary)) ){
if(VERBOSE) print("it's null")
if(VERBOSE) print(combinedAberrations$pathsummary)
tabout_drugNotAb = "Patient was not found to have aberrations in currently employed set of pathways."
if(VERBOSE) print("Patient was not found to have aberrations in currently employed set of pathways.")
}else{
tabout_drugNotAb = merge(x=coverage$pathsummary[drugNotAb,],
y=combinedAberrations$pathsummary,
by="path_id",
all.x=T,
sort=F)#merges so that all rows with drug screen summary will be included
if(VERBOSE) print(names(combinedAberrations))
if(VERBOSE) print("merged; zeroing the na's")
tabout_drugNotAb[is.na(tabout_drugNotAb[,11]),11]<- 0
hyperg_p_value_indexes = grep(pattern="hyperg_p_value",x=colnames(tabout_drugNotAb),ignore.case=T)
hyperg_p_value_index = hyperg_p_value_indexes[length(hyperg_p_value_indexes)]
#sort targeted, not aberrationally enriched by the proportion aberrational
tabout_drugNotAb = tabout_drugNotAb[order(tabout_drugNotAb[,hyperg_p_value_index],decreasing=F),]
}
outlist[["Drug targeted, not aberrationally enriched"]] = tabout_drugNotAb
allSensPathsMergAb = "No pathways were found to contain drug-sensitive genes"
if( !nullOrEmpty(functionalEnrichmentAnalysis$pathsummary) ){
allSensPathsMergAb = functionalEnrichmentAnalysis$pathsummary
if(VERBOSE) print("allSensPathsMergAb; merging to make allSensPathsMergAb")
if (is.null(dim(combinedAberrations$pathsummary)) ){
print("Patient was not found to have aberrations in currently employed set of pathways.")
}else{
tabout_allSensPathsMergAb = merge(x=functionalEnrichmentAnalysis$pathsummary,
y=combinedAberrations$pathsummary,
by="path_id",
all.x=T,
sort=F)#merges so that all rows with drug screen summary will be included
if(VERBOSE) print("merged; zeroing the na's")
tabout_allSensPathsMergAb[is.na(tabout_allSensPathsMergAb[,11]),11]<- 0
hyperg_p_value_indexes = grep(pattern="hyperg_p_value",x=colnames(tabout_allSensPathsMergAb),ignore.case=T)
hyperg_p_value_index = hyperg_p_value_indexes[length(hyperg_p_value_indexes)]
#sort targeted, not aberrationally enriched by the proportion aberrational
tabout_allSensPathsMergAb = tabout_allSensPathsMergAb[order(tabout_allSensPathsMergAb[,hyperg_p_value_index],decreasing=F),]
allSensPathsMergAb = tabout_allSensPathsMergAb
}
}
outlist[["Paths containing drug-sensitive genes"]] = allSensPathsMergAb
if(VERBOSE) print("sensenAndAb")
if( is.null(aben) ){
outlist[["Enriched for aberration and enriched for sensitive drug targets"]] = "Patient was not found to have aberrations in currently employed set of pathways."
outlist[["Aberration enriched, containing sensitive targets"]] = "Patient was not found to have aberrations in currently employed set of pathways."
}else{
sensenAndAb = intersect(sensen,aben)#enriched for sensitive targets and enriched for aberration
outlist[["Enriched for aberration and enriched for sensitive drug targets"]] = as.matrix(sensenAndAb,ncol=1)
if(VERBOSE) print("conSensAndAb")
conSensAndAb = intersect(senspaths,aben)#paths containing sensitive genes, that are also aberration enriched
outlist[["Aberration enriched, containing sensitive targets"]] = as.matrix(conSensAndAb, ncol=1)
}
# cat("\nWriting set of overlap-summary text files to:", paste("overlap_summary", as.character(Sys.time())), "\n")
# outHTMLname = paste(outfname, ".html",sep="")
#cat("\nWriting html-based summary to ", outHTMLname,"\n")
ret = list()
outlist[["combined_aberrations_summary"]] = combinedAberrations
outlist[["functional_drug_screen_summary"]] = functionalEnrichmentAnalysis
if(VERBOSE){
print(names(combinedAberrations))
print(names(combinedAberrations$path_summary_each_patient))
}
return(outlist)
}#coreOverlapAnalysis
#nullOrEmpty
#checks if an object, dat, is either null or empty, with zero rows or zero length
nullOrEmpty<-function(dat){
if(is.null(dat)) return(TRUE)
if(class(dat)=="matrix"|class(dat)=="data.frame"){
if(!nrow(dat)) return(TRUE)
}else{
if(!length(dat)) return(TRUE)
}
return(FALSE)
}
# olatest = abDrugOverlapAnalysis(results=results, path_detail=path_detail)
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.