diff CHM_Advanced.R @ 0:8893ea2915cc draft

Initial Version of Advanced Heat Map Tool
author insilico-bob
date Tue, 08 Aug 2017 14:01:05 -0400
parents
children 1f13d304ddbd
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/CHM_Advanced.R	Tue Aug 08 14:01:05 2017 -0400
@@ -0,0 +1,131 @@
+### This method generates a row and column ordering given an input matrix and ordering methods.
+###
+### matrixData - numeric matrix 
+### rowOrderMethod - Hierarchical, Original, Random
+### rowDistanceMeasure - For clustering, distance measure. May be: euclidean, binary, manhattan, maximum, canberra, minkowski, or correlation.
+### rowAgglomerationMethod - For clustering, agglomeration method.  May be:  'average' for Average Linkage, 'complete' for Complete Linkage,
+###                                                                          'single' for Single Linkage, 'ward', 'mcquitty', 'median', or 'centroid'.
+### colOrderMethod 
+### colDistanceMeasure
+### colAgglomerationMethod
+### rowOrderFile - output file of order of rows 
+### rowDendroFile - output file of row dendrogram  
+### colOrderFile - output file of order of cols
+### colDendroFile - output file of col dendrogram
+### rowCut - For rows the number of classifications to automatically generate based on dendrogram into a classification file.  0 for turned off.
+### colCut - For columns the number of classifications to automatically generate based on dendrogram into a classification file.  0 for turned off.
+
+performDataOrdering<-function(dataFile, rowOrderMethod, rowDistanceMeasure, rowAgglomerationMethod, colOrderMethod, colDistanceMeasure, colAgglomerationMethod,rowOrderFile, colOrderFile, rowDendroFile, colDendroFile, rowCut, colCut)
+{ 
+   dataMatrix = read.table(dataFile, header=TRUE, sep = "\t", row.names = 1, as.is=TRUE, na.strings=c("NA","N/A","-","?"))
+   rowOrder <-  createOrdering(dataMatrix, rowOrderMethod, "row", rowDistanceMeasure, rowAgglomerationMethod)  
+   if (rowOrderMethod == "Hierarchical") {
+      writeHCDataTSVs(rowOrder, rowDendroFile, rowOrderFile)
+	   if (rowCut != 0) {
+      		writeHCCut(rowOrder, rowCut, paste(rowOrderFile,".cut", sep=""))
+	   }
+   }
+
+   colOrder <-  createOrdering(dataMatrix, colOrderMethod, "col", colDistanceMeasure, colAgglomerationMethod)  
+   if (colOrderMethod == "Hierarchical") {
+      writeHCDataTSVs(colOrder, colDendroFile, colOrderFile)
+	   if (colCut != 0) {
+       		writeHCCut(colOrder, colCut, paste(colOrderFile,".cut", sep=""))
+	   }
+   }
+}
+
+#creates output files for hclust ordering
+writeHCDataTSVs<-function(uDend, outputHCDataFileName, outputHCOrderFileName)
+{
+   data<-cbind(uDend$merge, uDend$height, deparse.level=0)
+   colnames(data)<-c("A", "B", "Height")
+   write.table(data, file = outputHCDataFileName, append = FALSE, quote = FALSE, sep = "\t", row.names=FALSE)
+ 
+   data=matrix(,length(uDend$labels),2);
+   for (i in 1:length(uDend$labels)) {
+      data[i,1] = uDend$labels[i];
+      data[i,2] = which(uDend$order==i);
+   }
+   colnames(data)<-c("Id", "Order")
+   write.table(data, file = outputHCOrderFileName, append = FALSE, quote = FALSE, sep = "\t", row.names=FALSE)
+}
+
+#creates a classification file based on user specified cut of dendrogram
+writeHCCut<-function(uDend, cutNum, outputCutFileName)
+{
+   print (paste("Writing cut file ", outputCutFileName))
+   cut <- cutree(uDend, cutNum);
+   id <- names(cut);
+   data=matrix(,length(cut),2);
+   for (i in 1:length(cut)) {
+      data[i,1] = id[i];
+      data[i,2] = sprintf("Cluster %d", cut[i]);
+   }
+
+   write.table(data, file = outputCutFileName, append = FALSE, quote = FALSE, sep = "\t", row.names=FALSE, col.names = FALSE);
+}
+
+
+createOrdering<-function(matrixData, orderMethod, direction, distanceMeasure, agglomerationMethod)
+{
+  ordering <- NULL
+
+  if (orderMethod == "Hierarchical")
+  {
+
+    # Compute dendrogram for "Distance Metric"
+    distVals <- NULL
+    if(direction=="row") {
+      if (distanceMeasure == "correlation") {
+        geneGeneCor <- cor(t(matrixData), use="pairwise")
+        distVals <- as.dist((1-geneGeneCor)/2)
+      } else {
+        distVals <- dist(matrixData, method=distanceMeasure)
+      }
+    } else { #column
+      if (distanceMeasure == "correlation") {
+        geneGeneCor <- cor(matrixData, use="pairwise")
+        distVals <- as.dist((1-geneGeneCor)/2)
+      } else {
+        distVals <- dist(t(matrixData), method=distanceMeasure)
+      }
+    }
+
+#    if (agglomerationMethod == "ward") {
+#      ordering <- hclust(distVals * distVals, method="ward.D2")
+#    } else {
+      ordering <- hclust(distVals, method=agglomerationMethod)
+#    }
+  }
+  else if (orderMethod == "Random")
+  {
+    if(direction=="row") {
+       headerList <- rownames(matrixData)
+       ordering <- sample(headerList, length(headerList)) 
+    } else {
+       headerList <- colnames(matrixData)
+       ordering <- sample(headerList, length(headerList)) 
+    }
+  }
+  else if (orderMethod == "Original")
+  {
+    if(direction=="row") {
+       ordering <- rownames(matrixData) 
+    } else {
+       ordering <- colnames(matrixData) 
+    }
+  } else {
+    stop("createOrdering -- failed to find ordering method")
+  }
+  return(ordering)
+}
+### Initialize command line arguments and call performDataOrdering
+
+options(warn=-1)
+
+args = commandArgs(TRUE)
+
+performDataOrdering(dataFile=args[1], rowOrderMethod=args[2], rowDistanceMeasure=args[3], rowAgglomerationMethod=args[4], colOrderMethod=args[5], colDistanceMeasure=args[6], colAgglomerationMethod=args[7],rowOrderFile=args[8], colOrderFile=args[9], rowDendroFile=args[10], colDendroFile=args[11], rowCut=args[12], colCut=args[13])
+
+#suppressWarnings(performDataOrdering(dataFile=args[1], rowOrderMethod=args[2], rowDistanceMeasure=args[3], rowAgglomerationMethod=args[4], colOrderMethod=args[5], colDistanceMeasure=args[6], colAgglomerationMethod=args[7],rowOrderFile=args[8], colOrderFile=args[9], rowDendroFile=args[10], colDendroFile=args[11]))