Mercurial > repos > ethevenot > checkformat

comparison checkformat_script.R @ 1:e194eec8e70c draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/workflow4metabolomics/checkformat.git commit 8ebfbfa8d9449c9bbfbf569851a30b1e33df0b3f
author ethevenot
date Sat, 06 Aug 2016 11:54:28 -0400
parents
children 80a38d36f946
comparison
equal deleted inserted replaced
0:0d8099822c49 1:e194eec8e70c
1 ## Etienne Thevenot
2 ## CEA, MetaboHUB Paris
3 ## etienne.thevenot@cea.fr
4
5
6 ## Reads the dataMatrix, sampleMetadata, and variableMetadata .tsv files
7 ## and checks the formats
8 readAndCheckF <- function(datFilC="dataMatrix.tsv",
9 samFilC="sampleMetadata.tsv",
10 varFilC="variableMetadata.tsv") {
11
12 ## options
13
14 optStrAsFacL <- options()[["stringsAsFactors"]]
15 options(stringsAsFactors = FALSE)
16
17 ## checking that the tables have no duplicated row or column names
18
19 for(tabC in c("dat", "sam", "var")) {
20
21 tabNamC <- switch(tabC, dat="dataMatrix", sam="sampleMetadata", var="variableMetadata")
22
23 rowVc <- read.table(eval(parse(text=paste0(tabC, "FilC"))),
24 check.names = FALSE,
25 header = TRUE,
26 sep = "\t")[, 1]
27
28 colVc <- unlist(read.table(eval(parse(text=paste0(tabC, "FilC"))),
29 check.names = FALSE,
30 nrow=1,
31 sep = "\t"))[-1]
32
33 if(any(duplicated(rowVc)))
34 stop("The following row name(s) is/are duplicated in the ",
35 tabNamC,
36 " table: '",
37 paste(rowVc[duplicated(rowVc)], collapse="', '"), "'",
38 call.=FALSE)
39
40 if(any(duplicated(colVc)))
41 stop("The following column name(s) is/are duplicated in the ",
42 tabNamC,
43 " table: '",
44 paste(colVc[duplicated(colVc)], collapse="', '"), "'",
45 call.=FALSE)
46
47 rowMakVc <- make.names(rowVc, unique = TRUE)
48
49 rowDifVl <- rowVc != rowMakVc
50
51 if(any(rowDifVl)) {
52 rowDifDF <- data.frame(row = 1:length(rowVc),
53 actual = rowVc,
54 preferred = rowMakVc)
55 rowDifDF <- rowDifDF[rowDifVl, , drop = FALSE]
56 cat("\n\nWarning: The following row names of the ",
57 tabNamC,
58 " table are not in the standard R format, which may result in errors when loading the data in some of the W4M modules:\n", sep="")
59 print(rowDifDF)
60 }
61
62 colMakVc <- make.names(colVc, unique = TRUE)
63
64 colDifVl <- colVc != colMakVc
65
66 if(any(colDifVl)) {
67 colDifDF <- data.frame(col = 1:length(colVc),
68 actual = colVc,
69 preferred = colMakVc)
70 colDifDF <- colDifDF[colDifVl, , drop = FALSE]
71 cat("\n\nWarning: The following column names of the ",
72 tabNamC,
73 " table are not in the standard R format, which may result in errors when loading the data in some of the W4M modules:\n", sep="")
74 print(colDifDF)
75 }
76 }
77
78 ## reading tables
79
80 datMN <- t(as.matrix(read.table(datFilC,
81 check.names = FALSE,
82 header = TRUE,
83 row.names = 1,
84 sep = "\t")))
85
86 samDF <- read.table(samFilC,
87 check.names = FALSE,
88 header = TRUE,
89 row.names = 1,
90 sep = "\t")
91
92 varDF <- read.table(varFilC,
93 check.names = FALSE,
94 header = TRUE,
95 row.names = 1,
96 sep = "\t")
97
98 ## checking formats
99
100 chkL <- TRUE
101
102 if(!identical(rownames(datMN), rownames(samDF))) {
103 ## checking sample names
104
105 chkL <- FALSE
106
107 datSamDifVc <- setdiff(rownames(datMN), rownames(samDF))
108
109 if(length(datSamDifVc)) {
110 cat("\nThe following samples were found in the dataMatrix column names but not in the sampleMetadata row names:\n", sep="")
111 print(cbind.data.frame(col = as.numeric(sapply(datSamDifVc, function(samC) which(rownames(datMN) == samC))),
112 name = datSamDifVc))
113 }
114
115 samDatDifVc <- setdiff(rownames(samDF), rownames(datMN))
116
117 if(length(samDatDifVc)) {
118 cat("\n\nThe following samples were found in the sampleMetadata row names but not in the dataMatrix column names:\n", sep="")
119 print(cbind.data.frame(row = as.numeric(sapply(samDatDifVc, function(samC) which(rownames(samDF) == samC))),
120 name = samDatDifVc))
121 }
122
123 if(nrow(datMN) != nrow(samDF)) {
124 cat("\n\nThe dataMatrix has ", nrow(datMN), " columns (ie samples) whereas the sampleMetadata has ", nrow(samDF), " rows\n", sep="")
125 } else if(identical(gsub("^X", "", rownames(datMN)), rownames(samDF))) {
126 cat("\n\nThe dataMatrix column names start with an 'X' but not the sampleMetadata row names\n", sep="")
127 } else if(identical(gsub("^X", "", rownames(samDF)), rownames(datMN))) {
128 cat("\n\nThe sampleMetadata row names start with an 'X' but not the dataMatrix column names\n", sep="")
129 } else if(identical(sort(rownames(datMN)), sort(rownames(samDF)))) {
130 cat("\n\nThe dataMatrix column names and the sampleMetadata row names are not in the same order:\n", sep="")
131 print(cbind.data.frame(indice = 1:nrow(datMN),
132 dataMatrix_columnnames=rownames(datMN),
133 sampleMetadata_rownames=rownames(samDF))[rownames(datMN) != rownames(samDF), , drop = FALSE])
134 } else {
135 cat("\n\nThe dataMatrix column names and the sampleMetadata row names are not identical:\n", sep="")
136 print(cbind.data.frame(indice = 1:nrow(datMN),
137 dataMatrix_columnnames=rownames(datMN),
138 sampleMetadata_rownames=rownames(samDF))[rownames(datMN) != rownames(samDF), , drop = FALSE])
139 }
140
141 }
142
143 if(!identical(colnames(datMN), rownames(varDF))) {
144 ## checking variable names
145
146 chkL <- FALSE
147
148 datVarDifVc <- setdiff(colnames(datMN), rownames(varDF))
149
150 if(length(datVarDifVc)) {
151 cat("\nThe following variables were found in the dataMatrix row names but not in the variableMetadata row names:\n", sep="")
152 print(cbind.data.frame(row = as.numeric(sapply(datVarDifVc, function(varC) which(colnames(datMN) == varC))),
153 name = datVarDifVc))
154
155 }
156
157 varDatDifVc <- setdiff(rownames(varDF), colnames(datMN))
158
159 if(length(varDatDifVc)) {
160 cat("\n\nThe following variables were found in the variableMetadata row names but not in the dataMatrix row names:\n", sep="")
161 print(cbind.data.frame(row = as.numeric(sapply(varDatDifVc, function(varC) which(rownames(varDF) == varC))),
162 name = varDatDifVc))
163 }
164
165 if(ncol(datMN) != nrow(varDF)) {
166 cat("\n\nThe dataMatrix has ", nrow(datMN), " rows (ie variables) whereas the variableMetadata has ", nrow(varDF), " rows\n", sep="")
167 } else if(identical(sort(colnames(datMN)), sort(rownames(varDF)))) {
168 cat("\n\nThe dataMatrix row names and the variableMetadata row names are not in the same order:\n", sep="")
169 print(cbind.data.frame(row = 1:ncol(datMN),
170 dataMatrix_rownames=colnames(datMN),
171 variableMetadata_rownames=rownames(varDF))[colnames(datMN) != rownames(varDF), , drop = FALSE])
172 } else {
173 cat("\n\nThe dataMatrix row names and the variableMetadata row names are not identical:\n", sep="")
174 print(cbind.data.frame(row = 1:ncol(datMN),
175 dataMatrix_rownames=colnames(datMN),
176 variableMetadata_rownames=rownames(varDF))[colnames(datMN) != rownames(varDF), , drop = FALSE])
177 }
178 }
179
180 options(stringsAsFactors=optStrAsFacL)
181
182 resLs <- list(chkL=chkL)
183
184 return(resLs)
185
186 } ## end of checkAndReadF