code changes to support version 2 of the submission

R/4_Results_Benchmark.R

@@ -49,11 +49,12 @@ summaryDF <- melt(benchmarkDFPipeComp,
                                    "recallPeacoQC", "recallFlowAI"))
 
 summaryDF$pipeline <- ifelse(grepl("PeacoQC", summaryDF$indicator_pipe),
-                             "PeacoQC",
-                             "flowAI")
-summaryDF$pipeline <- factor(summaryDF$pipeline, levels = c("PeacoQC", "flowAI"))
+                             "PeacoQC-based",
+                             "flowAI-based")
+summaryDF$pipeline <- factor(summaryDF$pipeline, 
+                             levels = c("PeacoQC-based", "flowAI-based"))
 
-summaryDF$indicator <- ifelse(summaryDF$pipeline == "PeacoQC",
+summaryDF$indicator <- ifelse(summaryDF$pipeline == "PeacoQC-based",
                               gsub("PeacoQC", "", summaryDF$indicator_pipe),
                               gsub("FlowAI", "", summaryDF$indicator_pipe))
 
@@ -85,8 +86,14 @@ colnames(summaryDF2) <- c("dataset", "indicator", "PeacoQC", "flowAI")
 
 useRepel <- TRUE
 
+# combination of increased box.padding,
+# "" labels for non highlighted points,
+# and increased x and y coordinates, as advised in:
+# https://stackoverflow.com/questions/52397363/r-ggplot2-ggrepel-label-a-subset-of-points-while-being-aware-of-all-points
 geom_text_repel_high_overlap <- function(...){
-    geom_text_repel(..., max.overlaps = 100)
+    geom_text_repel(..., max.overlaps = 100, 
+                    box.padding = 1,
+                    min.segment.length = 0)
 }
 
 toDisplay <- c("D93_A05", "D91_C07", "D91_D03")
@@ -116,10 +123,10 @@ p <-
     geom_text_FUN(mapping = aes(label = label), hjust=0.5, vjust=1,
                   size = 2.5) +
     geom_abline(intercept = 0., slope = 1., colour = "blue", linetype = 2) +
-    coord_cartesian(xlim=c(0, 1), ylim=c(0, 1)) + 
+    coord_cartesian(xlim=c(-0.25, 1.25), ylim=c(-0.25, 1.25)) + 
     scale_colour_manual(values = c("false" = normalCol, 
                                    "true" = highlightCol)) +
-    xlab("PeacoQC pipeline") + ylab("flowAI pipeline") + 
+    xlab("PeacoQC-based pipeline") + ylab("flowAI-based pipeline") + 
     theme(legend.position="none") + 
     ggplot2::facet_wrap(~ indicator)
 p   

R/5_Discussion_NbOfEvents.R

@@ -0,0 +1,168 @@
+require(CytoPipeline)
+require(reshape2)
+require(ggplot2)
+require(patchwork)
+
+# Figure 14
+
+collectNbOfEvents <- function( 
+        experimentName,
+        path = ".",
+        whichSampleFiles) {
+
+    pipL <- CytoPipeline::buildCytoPipelineFromCache(
+        experimentName = experimentName,
+        path = path
+    )
+
+    if (missing(whichSampleFiles)) {
+        whichSampleFiles <- CytoPipeline::sampleFiles(pipL)
+    }
+
+    nEventPerSampleList <- list()
+    allStepNames <- c()
+    for (s in seq_along(whichSampleFiles)) {
+        message("Collecting nb of events for sample file ", whichSampleFiles[s], 
+                "...")
+        objInfos <- CytoPipeline::getCytoPipelineObjectInfos(
+            pipL,
+            whichQueue = "pre-processing",
+            sampleFile = whichSampleFiles[s])
+        objInfos <- objInfos[objInfos[,"ObjectClass"] == "flowFrame",]
+
+        nEventPerSampleList[[s]] <- lapply(
+            objInfos[,"ObjectName"],
+            FUN = function(objName) {
+                message("Reading object ", objName, "...")
+               ff <- CytoPipeline::getCytoPipelineFlowFrame(
+                   pipL,
+                   path = path,
+                   whichQueue = "pre-processing",
+                   sampleFile = whichSampleFiles[s],
+                   objectName = objName)
+               flowCore::nrow(ff)})
+
+        stepNames <- vapply(objInfos[,"ObjectName"],
+                            FUN = function(str){
+                                gsub(x = str,
+                                     pattern = "_obj",
+                                     replacement = "")
+                            },
+                            FUN.VALUE = character(length = 1))   
+
+        names(nEventPerSampleList[[s]]) <- stepNames
+
+        allStepNames <- union(allStepNames, stepNames)
+    }
+
+    nSampleFiles <- length(whichSampleFiles)
+    nAllSteps <- length(allStepNames)
+    eventNbs <- matrix(rep(NA, nSampleFiles * nAllSteps),
+                       nrow = nSampleFiles)
+    rownames(eventNbs) <- as.character(whichSampleFiles)
+    colnames(eventNbs) <- allStepNames
+
+    for (s in seq_along(whichSampleFiles)) {
+        stepNames <- names(nEventPerSampleList[[s]])
+        for (st in seq_along(stepNames)) {
+            eventNbs[as.character(whichSampleFiles)[s],
+                     stepNames[st]] <- 
+                nEventPerSampleList[[s]][[st]]
+        }
+    }
+
+    eventNbs
+
+}
+
+selectedExpName <- "HBVMouse_PQC"
+selectedSamples <- c("D91_C07.fcs", "D93_A05.fcs", "D91_D03.fcs")
+
+eventNbPeacoQC <- collectNbOfEvents( 
+    experimentName = selectedExpName,
+    whichSampleFiles = selectedSamples
+)
+
+eventFracPeacoQC <- t(apply(
+    eventNbPeacoQC,
+    MARGIN = 1,
+    FUN = function(line) {
+        if (length(line) == 0 || is.na(line[1])) {
+            as.numeric(rep(NA, length(line)))
+        } else {
+            line/line[1]
+        }
+    }
+))
+
+selectedExpName <- "HBVMouse_flowAI"
+
+eventNbFlowAI <- collectNbOfEvents( 
+    experimentName = selectedExpName,
+    whichSampleFiles = selectedSamples
+)
+
+eventFracFlowAI <- t(apply(
+    eventNbFlowAI,
+    MARGIN = 1,
+    FUN = function(line) {
+        if (length(line) == 0 || is.na(line[1])) {
+            as.numeric(rep(NA, length(line)))
+        } else {
+            line/line[1]
+        }
+    }
+))
+
+stepNames <- colnames(eventFracPeacoQC)
+DFPeacoQC <- as.data.frame(eventFracPeacoQC)
+DFPeacoQC$sampleFile <- rownames(DFPeacoQC)
+DFPeacoQC2 <- reshape(data = DFPeacoQC,
+               direction = "long",
+               v.names = "eventFrac",
+               varying = stepNames, 
+               timevar = "step",
+               times = stepNames)
+
+DFPeacoQC2$step <- factor(DFPeacoQC2$step,
+                          levels = stepNames)
+
+p1 <- ggplot(DFPeacoQC2,
+             mapping = aes(x = step,
+                           y = eventFrac,
+                           group = sampleFile,
+                           col = sampleFile)) +
+    geom_line() +
+    scale_y_continuous(limits=c(0,1)) + 
+    theme(axis.text.x = element_text(angle = 90),
+          legend.position = "none") +
+    labs(y = "fraction of events kept", 
+         title = "PeacoQC-based pipeline")
+
+stepNames <- colnames(eventFracFlowAI)
+DFFlowAI <- as.data.frame(eventFracFlowAI)
+DFFlowAI$sampleFile <- rownames(DFFlowAI)
+DFFlowAI2 <- reshape(data = DFFlowAI,
+                      direction = "long",
+                      v.names = "eventFrac",
+                      varying = stepNames, 
+                      timevar = "step",
+                      times = stepNames)
+
+DFFlowAI2$step <- factor(DFFlowAI2$step,
+                          levels = stepNames)
+
+p2 <- ggplot(DFFlowAI2,
+             mapping = aes(x = step,
+                           y = eventFrac,
+                           group = sampleFile,
+                           col = sampleFile)) +
+    geom_line() +
+    scale_y_continuous(limits=c(0,1)) + 
+    theme(axis.text.x = element_text(angle = 90)) +
+    labs(y = "fraction of events kept", 
+         title = "flowAI-based pipeline")
+
+
+p1+p2
+