topicsRcisTarget error about cistopic

[cbravo93]

Hi!

I have tried it with feather 0.3.3 and RcisTarget (from github) 1.5.0 and it seems to work. Can you check if it works for you with this:

topicsRcisTarget <- function(
  object,
  genome,
  pathToDb,
  reduced_database = FALSE,
  nesThreshold = 3,
  rocthr = 0.005,
  maxRank = 20000,
  nCores = 1,
  ...
){
  # Check input
  if(length([e-mail]) < 1){
    stop('Please, run binarizedcisTopicsToCtx() first.')
  }
  
  # Check dependencies
  if(! "RcisTarget" %in% installed.packages()){
    stop('Please, install RcisTaregt: \n install_github("aertslab/RcisTarget")')
  } else {
    require(RcisTarget)
  }
  
  if (genome == 'hg19'){
    data(motifAnnotations_hgnc)
    motifAnnot <- motifAnnotations_hgnc
    if(rocthr!=0.005 | maxRank!=20000){
      warning("For Homo sapiens the recommended settings are: rocthr=0.005, maxRank=20000")
    } 
  }
  else if (genome == 'mm9'){
    data(motifAnnotations_mgi)
    motifAnnot <- motifAnnotations_mgi
    if(rocthr!=0.005 | maxRank!=20000){
      warning("For Mus musculus the recommended settings are: rocthr=0.005, maxRank=20000")
    } 
  }
  else if (genome == 'dm3'){
    data(motifAnnotations_dmel)
    motifAnnot <- motifAnnotations_dmel
    if(rocthr!=0.01 | maxRank!=5000){
      warning("For Drosophila melanogaster the recommended settings are: rocthr=0.01, maxRank=5000")
    } 
  }
  else if (genome == 'dm6'){
    data(motifAnnotations_dmel)
    motifAnnot <- motifAnnotations_dmel
    if(rocthr!=0.01 | maxRank!=5000){
      warning("For Drosophila melanogaster the recommended settings are: rocthr=0.01, maxRank=5000")
    } 
  } else {
    stop('The genome required is not available! Try using the liftover option.')
  }

  topicsList <- [e-mail]
  
  extension <- strsplit(pathToDb, "\\.")[[1]][length(strsplit(pathToDb, "\\.")[[1]])]
  if (extension == 'feather'){
    columnsinRanking <- feather::feather_metadata(pathToDb)[["dim"]][2]-1
  }
  else if (extension == "parquet"){
    pq <- arrow::parquet_file_reader(pathToDb)
    columnsinRanking <- pq$GetSchema()$num_fields()-1
  }
  else{
    stop("Database format must be feather or parquet.")
  }

  if (reduced_database == FALSE){
    ctxreg <- unique(as.vector(unlist([e-mail])))
    motifRankings <- importRankings(pathToDb, columns = c('features', ctxreg))
  }
  else{
    motifRankings <- importRankings(pathToDb)
    ctxregions <- colnames(getRanking(motifRankings))[-1]
    topicsList <- llply(1:length(topicsList), function(i) topicsList[[i]][which(topicsList[[i]] %in% ctxregions)])
    names(topicsList) <- names([e-mail])
  }
  
  if (length(topicsList) < nCores){
    print(paste('The number of cores (', nCores, ') is higher than the number of topics (', length(topicsList),').', sep=''))
  }

  if(nCores > 1){
    cl <- makeCluster(nCores, type = "SOCK")
    registerDoSNOW(cl)
    print(paste('Exporting data to clusters...'))
    clusterEvalQ(cl, library(feather), library(RcisTarget))
    clusterExport(cl, c("topicsList", "motifRankings", "motifAnnot", "nesThreshold", "rocthr", "columnsinRanking", "maxRank"), envir=environment())
    print(paste('Running RcisTarget...'))
    cisTopic.cisTarget <- suppressWarnings(llply(1:length(topicsList), function (i) cisTarget(topicsList[[i]],
                                    motifRankings,
                                    motifAnnot = motifAnnot,
                                    nesThreshold = nesThreshold,
                                    aucMaxRank = rocthr * columnsinRanking,
                                    geneErnMmaxRank = maxRank,
                                    nCores=1
    ), .parallel = TRUE))
    stopCluster(cl)
  }
  else{
    cisTopic.cisTarget <- suppressWarnings(llply(1:length(topicsList), function (i) cisTarget(topicsList[[i]],
                                                                                             motifRankings,
                                                                                             motifAnnot = motifAnnot,
                                                                                             nesThreshold = nesThreshold,
                                                                                             aucMaxRank = rocthr * columnsinRanking,
                                                                                             geneErnMmaxRank = maxRank,
                                                                                             nCores=1
    )))
  }

  object.binarized.RcisTarget <- list()

  for (i in 1:length(cisTopic.cisTarget)){
    if(nrow(cisTopic.cisTarget[[i]]) > 0){
      colnames(cisTopic.cisTarget[[i]])[c(1, 7, 9)] <- c('cisTopic', 'nEnrRegions', 'enrichedRegions')
      cisTopic.cisTarget[[i]]$cisTopic <- rep(paste('Topic', i, sep='_'), nrow(cisTopic.cisTarget[[i]]))
      object.binarized.RcisTarget[[i]] <- addLogo(cisTopic.cisTarget[[i]])
    } else {
      cisTopic.cisTarget[[i]] <- NULL
    }
    
  }

  [e-mail] <- object.binarized.RcisTarget
  return(object)
}

I would also lower the number of cores a bit, because it will load the feather database in each of them (and that can take up a lot of memory).

Let me know if this solves it!

C

[jk86754]

Unfortunately, same issue. I am wondering if i am having a docker-related issue. My RStudio is running on a docker container and i am starting to think this is causing the issue. I can run the command with ncores=1 (although it does take forever); the error comes up with ncores>1. As far as i recall, i had to do the same thing for SCENIC.