Mercurial > repos > md-anderson-bioinformatics > heat_map_creation_advanced

comparison CHM_Advanced.R @ 0:8893ea2915cc draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Initial Version of Advanced Heat Map Tool
author insilico-bob
date Tue, 08 Aug 2017 14:01:05 -0400
parents
children 1f13d304ddbd
comparison
equal deleted inserted replaced
-1:000000000000 0:8893ea2915cc
1 ### This method generates a row and column ordering given an input matrix and ordering methods.
2 ###
3 ### matrixData - numeric matrix
4 ### rowOrderMethod - Hierarchical, Original, Random
5 ### rowDistanceMeasure - For clustering, distance measure. May be: euclidean, binary, manhattan, maximum, canberra, minkowski, or correlation.
6 ### rowAgglomerationMethod - For clustering, agglomeration method. May be: 'average' for Average Linkage, 'complete' for Complete Linkage,
7 ### 'single' for Single Linkage, 'ward', 'mcquitty', 'median', or 'centroid'.
8 ### colOrderMethod
9 ### colDistanceMeasure
10 ### colAgglomerationMethod
11 ### rowOrderFile - output file of order of rows
12 ### rowDendroFile - output file of row dendrogram
13 ### colOrderFile - output file of order of cols
14 ### colDendroFile - output file of col dendrogram
15 ### rowCut - For rows the number of classifications to automatically generate based on dendrogram into a classification file. 0 for turned off.
16 ### colCut - For columns the number of classifications to automatically generate based on dendrogram into a classification file. 0 for turned off.
17
18 performDataOrdering<-function(dataFile, rowOrderMethod, rowDistanceMeasure, rowAgglomerationMethod, colOrderMethod, colDistanceMeasure, colAgglomerationMethod,rowOrderFile, colOrderFile, rowDendroFile, colDendroFile, rowCut, colCut)
19 {
20 dataMatrix = read.table(dataFile, header=TRUE, sep = "\t", row.names = 1, as.is=TRUE, na.strings=c("NA","N/A","-","?"))
21 rowOrder <- createOrdering(dataMatrix, rowOrderMethod, "row", rowDistanceMeasure, rowAgglomerationMethod)
22 if (rowOrderMethod == "Hierarchical") {
23 writeHCDataTSVs(rowOrder, rowDendroFile, rowOrderFile)
24 if (rowCut != 0) {
25 writeHCCut(rowOrder, rowCut, paste(rowOrderFile,".cut", sep=""))
26 }
27 }
28
29 colOrder <- createOrdering(dataMatrix, colOrderMethod, "col", colDistanceMeasure, colAgglomerationMethod)
30 if (colOrderMethod == "Hierarchical") {
31 writeHCDataTSVs(colOrder, colDendroFile, colOrderFile)
32 if (colCut != 0) {
33 writeHCCut(colOrder, colCut, paste(colOrderFile,".cut", sep=""))
34 }
35 }
36 }
37
38 #creates output files for hclust ordering
39 writeHCDataTSVs<-function(uDend, outputHCDataFileName, outputHCOrderFileName)
40 {
41 data<-cbind(uDend$merge, uDend$height, deparse.level=0)
42 colnames(data)<-c("A", "B", "Height")
43 write.table(data, file = outputHCDataFileName, append = FALSE, quote = FALSE, sep = "\t", row.names=FALSE)
44
45 data=matrix(,length(uDend$labels),2);
46 for (i in 1:length(uDend$labels)) {
47 data[i,1] = uDend$labels[i];
48 data[i,2] = which(uDend$order==i);
49 }
50 colnames(data)<-c("Id", "Order")
51 write.table(data, file = outputHCOrderFileName, append = FALSE, quote = FALSE, sep = "\t", row.names=FALSE)
52 }
53
54 #creates a classification file based on user specified cut of dendrogram
55 writeHCCut<-function(uDend, cutNum, outputCutFileName)
56 {
57 print (paste("Writing cut file ", outputCutFileName))
58 cut <- cutree(uDend, cutNum);
59 id <- names(cut);
60 data=matrix(,length(cut),2);
61 for (i in 1:length(cut)) {
62 data[i,1] = id[i];
63 data[i,2] = sprintf("Cluster %d", cut[i]);
64 }
65
66 write.table(data, file = outputCutFileName, append = FALSE, quote = FALSE, sep = "\t", row.names=FALSE, col.names = FALSE);
67 }
68
69
70 createOrdering<-function(matrixData, orderMethod, direction, distanceMeasure, agglomerationMethod)
71 {
72 ordering <- NULL
73
74 if (orderMethod == "Hierarchical")
75 {
76
77 # Compute dendrogram for "Distance Metric"
78 distVals <- NULL
79 if(direction=="row") {
80 if (distanceMeasure == "correlation") {
81 geneGeneCor <- cor(t(matrixData), use="pairwise")
82 distVals <- as.dist((1-geneGeneCor)/2)
83 } else {
84 distVals <- dist(matrixData, method=distanceMeasure)
85 }
86 } else { #column
87 if (distanceMeasure == "correlation") {
88 geneGeneCor <- cor(matrixData, use="pairwise")
89 distVals <- as.dist((1-geneGeneCor)/2)
90 } else {
91 distVals <- dist(t(matrixData), method=distanceMeasure)
92 }
93 }
94
95 # if (agglomerationMethod == "ward") {
96 # ordering <- hclust(distVals * distVals, method="ward.D2")
97 # } else {
98 ordering <- hclust(distVals, method=agglomerationMethod)
99 # }
100 }
101 else if (orderMethod == "Random")
102 {
103 if(direction=="row") {
104 headerList <- rownames(matrixData)
105 ordering <- sample(headerList, length(headerList))
106 } else {
107 headerList <- colnames(matrixData)
108 ordering <- sample(headerList, length(headerList))
109 }
110 }
111 else if (orderMethod == "Original")
112 {
113 if(direction=="row") {
114 ordering <- rownames(matrixData)
115 } else {
116 ordering <- colnames(matrixData)
117 }
118 } else {
119 stop("createOrdering -- failed to find ordering method")
120 }
121 return(ordering)
122 }
123 ### Initialize command line arguments and call performDataOrdering
124
125 options(warn=-1)
126
127 args = commandArgs(TRUE)
128
129 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])
130
131 #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]))