update release_1.0.3

GreenleafLab · Jun 3, 2022 · 55f0923 · 55f0923
1 parent 2b3c722
commit 55f0923
Show file tree

Hide file tree

Showing 17 changed files with 262 additions and 70 deletions.
diff --git a/.DS_Store b/.DS_Store
diff --git a/NAMESPACE b/NAMESPACE
@@ -15,6 +15,7 @@ export(addArchRAnnotations)
 export(addArchRChrPrefix)
 export(addArchRDebugging)
 export(addArchRGenome)
+export(addArchRLocking)
 export(addArchRLogging)
 export(addArchRThreads)
 export(addArchRVerbose)
@@ -153,6 +154,7 @@ export(projectBulkATAC)
 export(recoverArchRProject)
 export(reformatFragmentFiles)
 export(saveArchRProject)
+export(setArchRLocking)
 export(subsetArchRProject)
 export(subsetCells)
 export(theme_ArchR)

diff --git a/R/AnnotationPeaks.R b/R/AnnotationPeaks.R
@@ -390,41 +390,57 @@ addMotifAnnotations <- function(
     motifs <- obj$motifs
     motifSummary <- obj$motifSummary
 
-  }else if(tolower(motifSet)=="cisbp"){
+  }else if(tolower(motifSet) %in% c("cisbp", "cisbptest")){
 
     .requirePackage("chromVARmotifs",installInfo='devtools::install_github("GreenleafLab/chromVARmotifs")')
-    if(tolower(species) == "mus musculus"){
-      if(version == 1){
-        message("Using version 1 motifs!")
-        data("mouse_pwms_v1")
-        motifs <- mouse_pwms_v1        
-      }else if(version == 2){
-        message("Using version 2 motifs!")
-        data("mouse_pwms_v2")
-        motifs <- mouse_pwms_v2
-      }else{
-        stop("Only versions 1 and 2 exist!")
-      }
-      obj <- .summarizeChromVARMotifs(motifs)
-      motifs <- obj$motifs
-      motifSummary <- obj$motifSummary
-    }else if(tolower(species) == "homo sapiens"){
-      if(version == 1){
-        message("Using version 1 motifs!")
-        data("human_pwms_v1")
-        motifs <- human_pwms_v1        
-      }else if(version == 2){
+
+    if(tolower(motifSet) == "cisbptest"){
+
         message("Using version 2 motifs!")
         data("human_pwms_v2")
         motifs <- human_pwms_v2
+        subset <- grep("PAX5|CEBPA|CEBPB|IRF4|ETS1|EOMES", names(motifs), value=TRUE)
+        motifs <- motifs[subset]
+        obj <- .summarizeChromVARMotifs(motifs)
+        motifs <- obj$motifs
+        motifSummary <- obj$motifSummary
+
+    }else{
+
+      if(tolower(species) == "mus musculus"){
+        if(version == 1){
+          message("Using version 1 motifs!")
+          data("mouse_pwms_v1")
+          motifs <- mouse_pwms_v1        
+        }else if(version == 2){
+          message("Using version 2 motifs!")
+          data("mouse_pwms_v2")
+          motifs <- mouse_pwms_v2
+        }else{
+          stop("Only versions 1 and 2 exist!")
+        }
+        obj <- .summarizeChromVARMotifs(motifs)
+        motifs <- obj$motifs
+        motifSummary <- obj$motifSummary
+      }else if(tolower(species) == "homo sapiens"){
+        if(version == 1){
+          message("Using version 1 motifs!")
+          data("human_pwms_v1")
+          motifs <- human_pwms_v1        
+        }else if(version == 2){
+          message("Using version 2 motifs!")
+          data("human_pwms_v2")
+          motifs <- human_pwms_v2
+        }else{
+          stop("Only versions 1 and 2 exist!")
+        }
+        obj <- .summarizeChromVARMotifs(motifs)
+        motifs <- obj$motifs
+        motifSummary <- obj$motifSummary
       }else{
-        stop("Only versions 1 and 2 exist!")
+        stop("Species not recognized homo sapiens, mus musculus supported by CisBP!")
       }
-      obj <- .summarizeChromVARMotifs(motifs)
-      motifs <- obj$motifs
-      motifSummary <- obj$motifSummary
-    }else{
-      stop("Species not recognized homo sapiens, mus musculus supported by CisBP!")
+
     }
 
   }else if(tolower(motifSet)=="encode"){

diff --git a/R/CreateArrow.R b/R/CreateArrow.R
@@ -202,12 +202,21 @@ createArrowFiles <- function(
   args$registryDir <- file.path(QCDir, "CreateArrowsRegistry")
   args$cleanTmp <- NULL
 
-  if(subThreading){
-    h5disableFileLocking()
-  }else{
+  #H5 File Lock Check
+  h5lock <- setArchRLocking()
+  if(h5lock){
+    if(subThreading){
+      message("subThreadhing Disabled since ArchRLocking is TRUE see `addArchRLocking`")
+      subThreading <- FALSE
+    }
     args$threads <- length(inputFiles)
+  }else{
+    if(subThreading){
+      message("subThreadhing Enabled since ArchRLocking is FALSE see `addArchRLocking`")
+    }    
   }
 
+  #Default Param
   args$minTSS <- NULL
 
   #Run With Parallel or lapply
@@ -228,10 +237,6 @@ createArrowFiles <- function(
     paste0(args$outputNames,".arrow")[file.exists(paste0(args$outputNames,".arrow"))]
   })
 
-  if(subThreading){
-    h5enableFileLocking()
-  }
-
   .endLogging(logFile = logFile)
 
   return(outArrows)

diff --git a/R/GlobalDefaults.R b/R/GlobalDefaults.R
@@ -5,6 +5,7 @@
 ########################################################
 ArchRDefaults <- list(
   ArchR.threads = 1,
+  ArchR.locking = FALSE,
   ArchR.logging = TRUE,
   ArchR.genome = NA,
   ArchR.chrPrefix = TRUE,
@@ -359,6 +360,80 @@ getArchRThreads <- function(){
   }
 }
 
+##########################################################################################
+# H5 File Locking
+##########################################################################################
+
+#' Add a globally-applied H5 file locking setup
+#' 
+#' This function will set the default H5 file locking parameters
+#' 
+#' @param locking The default value for H5 File Locking
+#' @export
+addArchRLocking <- function(locking=FALSE){
+
+  .validInput(input = locking, name = "locking", valid = "boolean")
+
+  #Check if Lockign is Valid
+  h5test <- h5testFileLocking(".")
+  if(!h5test){
+    message(
+      "H5 Locking is not enabled based on 'h5testFileLocking'.\nSetting ArchRLocking locking to ", 
+      locking, "."
+    )
+    locking <- TRUE
+  }else{
+    message("Setting ArchRLocking to ", locking, ".")
+  }
+  options(ArchR.locking = locking)
+
+}
+
+#' Set a globally-applied H5 file locking setup
+#' 
+#' This function will set the default H5 file locking parameters to the system
+#' 
+#' @export
+setArchRLocking <- function(){
+
+  #Get Value
+  .ArchRLocking <- options()[["ArchR.locking"]]
+  if(is.null(.ArchRLocking)){
+    .ArchRLocking <- TRUE
+  }else if(!is.logical(.ArchRLocking)){
+    .ArchRLocking <- TRUE
+  }
+
+  #Get Environment Value
+  h5lock <- tryCatch({
+      Sys.getenv("HDF5_USE_FILE_LOCKING")
+    }, error = function(e){
+      ""
+  })
+  if(h5lock=="FALSE"){
+    h5lock <- FALSE
+  }else if(h5lock==""){
+    h5lock <- TRUE
+  }else{
+    stop("H5 Locking Not Valid!")
+  }
+
+  #Set Environmental Value
+  if(.ArchRLocking != h5lock){
+    if(.ArchRLocking){
+      message("Enabling H5 File Locking. If this is not desired check `addArchRLocking`.")
+      h5enableFileLocking()
+    }else{
+      message("Disabling H5 File Locking. If this is not desired check `addArchRLocking`.")
+      h5disableFileLocking()
+    }
+  }
+
+  #Return Value
+  .ArchRLocking
+
+}
+
 ##########################################################################################
 # Create Gene/Genome Annotation
 ##########################################################################################
@@ -380,7 +455,7 @@ addArchRGenome <- function(genome = NULL, install = TRUE){
   .validInput(input = genome, name = "genome", valid = "character")
   .validInput(input = install, name = "install", valid = c("boolean"))
 
-  supportedGenomes <- c("hg19","hg38","mm9","mm10","hg19test")
+  supportedGenomes <- c("hg19","hg38","mm9","mm10","hg19test", "hg19test2")
 
   if(tolower(genome) %ni% supportedGenomes){
 
@@ -400,7 +475,7 @@ addArchRGenome <- function(genome = NULL, install = TRUE){
           stop("BSgenome for hg19 not installed! Please install by setting install = TRUE or by the following:\n\tBiocManager::install(\"BSgenome.Hsapiens.UCSC.hg19\")")
         }
       }
-    }else if(tolower(genome)=="hg19test"){
+    }else if(tolower(genome) %in% c("hg19test", "hg19test2")){
       if(!requireNamespace("BSgenome.Hsapiens.UCSC.hg19", quietly = TRUE)){
         if(install){
           message("BSgenome for hg19 not installed! Now installing by the following:\n\tBiocManager::install(\"BSgenome.Hsapiens.UCSC.hg19\")")
@@ -470,7 +545,7 @@ getArchRGenome <- function(
   .validInput(input = geneAnnotation, name = "geneAnnotation", valid = "boolean")
   .validInput(input = genomeAnnotation, name = "genomeAnnotation", valid = "boolean")
 
-  supportedGenomes <- c("hg19","hg38","mm9","mm10","hg19test")
+  supportedGenomes <- c("hg19","hg38","mm9","mm10","hg19test", "hg19test2")
   .ArchRGenome <- options()[["ArchR.genome"]]
 
   if(!is.null(.ArchRGenome)){

diff --git a/R/GroupCoverages.R b/R/GroupCoverages.R
@@ -212,19 +212,26 @@ addGroupCoverages <- function(
   )
   args$covDir <- file.path(getOutputDirectory(ArchRProj), "GroupCoverages", groupBy)
   args$parallelParam <- parallelParam
-  args$threads <- threads
   args$verbose <- verbose
   args$tstart <- tstart
   args$logFile <- logFile
   args$registryDir <- file.path(getOutputDirectory(ArchRProj), "GroupCoverages", "batchRegistry")
 
+  #H5 File Lock Check
+  h5lock <- setArchRLocking()
+  if(h5lock){
+    args$threads <- 1
+  }else{
+    if(threads > 1){
+      message("subThreadhing Enabled since ArchRLocking is FALSE see `addArchRLocking`")
+    }
+    args$threads <- threads
+  }
+
   #####################################################
   # Batch Apply to Create Insertion Coverage Files
   #####################################################
 
-  #Disable Hdf5 File Locking
-  h5disableFileLocking()
-
   #Batch Apply
   .logDiffTime(sprintf("Creating Coverage Files!"), tstart, addHeader = FALSE)
   batchOut <- .batchlapply(args)
@@ -256,9 +263,6 @@ addGroupCoverages <- function(
 
   ArchRProj@projectMetadata$GroupCoverages[[groupBy]] <- SimpleList(Params = Params, coverageMetadata = coverageMetadata)
 
-  #Enable Hdf5 File Locking
-  h5enableFileLocking()
-
   .logDiffTime(sprintf("Finished Creation of Coverage Files!"), tstart, addHeader = FALSE)
   .endLogging(logFile = logFile)
 

diff --git a/R/GroupExport.R b/R/GroupExport.R
@@ -233,8 +233,15 @@ getGroupBW <- function(
   chromSizes <- getChromSizes(ArchRProj)
   tiles <- unlist(slidingWindows(chromSizes, width = tileSize, step = tileSize))
 
-  if(threads > 1){
-    h5disableFileLocking()
+  #H5 File Lock Check
+  h5lock <- setArchRLocking()
+  if(h5lock){
+    threads <- 1
+  }else{
+    if(threads > 1){
+      message("subThreadhing Enabled since ArchRLocking is FALSE see `addArchRLocking`")
+    }
+    args$threads <- threads
   }
 
   covFiles <- c()
@@ -264,10 +271,6 @@ getGroupBW <- function(
 
   }
 
-  if(threads > 1){
-    h5enableFileLocking()
-  }
-
   .endLogging(logFile = logFile)
 
   covFiles

diff --git a/R/IterativeLSI.R b/R/IterativeLSI.R
@@ -144,6 +144,20 @@ addIterativeLSI <- function(
     stop("Please provide more than 1000 varFeatures!")
   }
 
+  if(nCells(ArchRProj) < 500){
+     message(
+        "Detected less than 500 Cells.\n",
+        "\t`filterBias` disabled.\n",
+        "\t`outlierQuantiles` disabled\n",
+        "\t`sampleCellsPre` disabled\n",
+        "\t`testBias` in `addClusters` disabled\n"
+      )
+     filterBias <- FALSE
+     outlierQuantiles <- c(0, 1)
+     sampleCellsPre <- NULL
+     clusterParams$testBias <- FALSE
+  }
+
   .startLogging(logFile = logFile)
   .logThis(mget(names(formals()),sys.frame(sys.nframe())), "IterativeLSI Input-Parameters", logFile=logFile)
 
@@ -235,8 +249,14 @@ addIterativeLSI <- function(
     .logDiffTime("Computing Top Features", tstart, addHeader = FALSE, verbose = verbose, logFile = logFile)
     nFeature <- varFeatures[1]
     rmTop <- floor((1-filterQuantile) * totalFeatures)
-    topIdx <- head(order(totalAcc$rowSums, decreasing=TRUE), nFeature + rmTop)[-seq_len(rmTop)]
+    if(sum(totalAcc$rowSums > 0) > 2.25 * varFeatures){
+      topIdx <- head(order(totalAcc$rowSums, decreasing=TRUE), nFeature + rmTop)[-seq_len(rmTop)]
+    }else{
+      message("Not Enough Non-Zero Features to Filter!")
+      topIdx <- head(order(totalAcc$rowSums, decreasing=TRUE), nFeature)
+    }
     topFeatures <- totalAcc[sort(topIdx),]
+    topFeatures <- topFeatures[topFeatures$rowSums > 0,]
 
     gc()
 
@@ -268,10 +288,11 @@ addIterativeLSI <- function(
     .logDiffTime("Computing Variable Features", tstart, addHeader = FALSE, verbose = verbose, logFile = logFile)
     nFeature <- varFeatures[1]
     if(nFeature > 0.5 * nrow(totalAcc)){
-      stop("nFeature for variable selection must be at leat 1/2 the total features!")
+      stop("nFeature for variable selection must be at most 1/2 the total features!")
     }
     topIdx <- head(order(totalAcc$combinedVars, decreasing=TRUE), nFeature)
     topFeatures <- totalAcc[sort(topIdx),]
+    topFeatures <- topFeatures[topFeatures$combinedMeans > 0,]
 
     gc()
 
@@ -281,6 +302,10 @@ addIterativeLSI <- function(
 
   }
 
+  if(nrow(topFeatures) < varFeatures){
+    stop(sprintf("Not Enough Features Found in data (%s)!", nrow(topFeatures)))
+  }
+
   cellDepth <- tryCatch({
       df <- getCellColData(ArchRProj = ArchRProj, select = depthCol)
       v <- df[,1]