Mercurial > repos > ethevenot > checkformat
comparison checkformat_script.R @ 4:9590fac86f63 draft
planemo upload for repository https://github.com/workflow4metabolomics/checkformat.git commit 46e92656192f2714265cc525b283c7cbb87b9093
| author | ethevenot |
|---|---|
| date | Wed, 28 Feb 2018 05:47:44 -0500 |
| parents | 80a38d36f946 |
| children |
comparison
equal
deleted
inserted
replaced
| 3:80a38d36f946 | 4:9590fac86f63 |
|---|---|
| 80 } | 80 } |
| 81 | 81 |
| 82 if(nrow(datMN) != nrow(samDF)) { | 82 if(nrow(datMN) != nrow(samDF)) { |
| 83 if(nrow(datMN) > nrow(samDF)) { | 83 if(nrow(datMN) > nrow(samDF)) { |
| 84 print(setdiff(rownames(datMN), rownames(samDF))) | 84 print(setdiff(rownames(datMN), rownames(samDF))) |
| 85 stop("The sample names above from dataMatrix were not found in sampleMetadata") | 85 stop("The sample names above from dataMatrix were not found in sampleMetadata", |
| 86 call. = FALSE) | |
| 86 } else { | 87 } else { |
| 87 print(setdiff(rownames(samDF), rownames(datMN))) | 88 print(setdiff(rownames(samDF), rownames(datMN))) |
| 88 stop("The sample names above from sampleMetadata were not found in dataMatrix") | 89 stop("The sample names above from sampleMetadata were not found in dataMatrix", |
| 90 call. = FALSE) | |
| 89 } | 91 } |
| 90 } | 92 } |
| 91 | 93 |
| 92 if(ncol(datMN) != nrow(varDF)) { | 94 if(ncol(datMN) != nrow(varDF)) { |
| 93 if(ncol(datMN) > nrow(varDF)) { | 95 if(ncol(datMN) > nrow(varDF)) { |
| 94 print(setdiff(colnames(datMN), rownames(varDF))) | 96 print(setdiff(colnames(datMN), rownames(varDF))) |
| 95 stop("The variable names above from dataMatrix were not found in variableMetadata") | 97 stop("The variable names above from dataMatrix were not found in variableMetadata", |
| 98 call. = FALSE) | |
| 96 } else { | 99 } else { |
| 97 print(setdiff(rownames(varDF), colnames(datMN))) | 100 print(setdiff(rownames(varDF), colnames(datMN))) |
| 98 stop("The variable names above from variableMetadata were not found in dataMatrix") | 101 stop("The variable names above from variableMetadata were not found in dataMatrix", |
| 102 call. = FALSE) | |
| 99 } | 103 } |
| 100 } | 104 } |
| 101 | 105 |
| 102 | 106 |
| 103 ## making sample and variable names (optional) | 107 ## making sample and variable names (optional) |
| 179 | 183 |
| 180 return(resLs) | 184 return(resLs) |
| 181 | 185 |
| 182 } ## end of checkAndReadF | 186 } ## end of checkAndReadF |
| 183 | 187 |
| 184 | |
| 185 | |
| 186 | |
| 187 ## if(!identical(rownames(datMN), rownames(samDF))) { | |
| 188 ## ## checking sample names | |
| 189 | |
| 190 ## chkL <- FALSE | |
| 191 | |
| 192 ## datSamDifVc <- setdiff(rownames(datMN), rownames(samDF)) | |
| 193 | |
| 194 ## if(length(datSamDifVc)) { | |
| 195 ## cat("\nThe following samples were found in the dataMatrix column names but not in the sampleMetadata row names:\n", sep="") | |
| 196 ## print(cbind.data.frame(col = as.numeric(sapply(datSamDifVc, function(samC) which(rownames(datMN) == samC))), | |
| 197 ## name = datSamDifVc)) | |
| 198 ## } | |
| 199 | |
| 200 ## samDatDifVc <- setdiff(rownames(samDF), rownames(datMN)) | |
| 201 | |
| 202 ## if(length(samDatDifVc)) { | |
| 203 ## cat("\n\nThe following samples were found in the sampleMetadata row names but not in the dataMatrix column names:\n", sep="") | |
| 204 ## print(cbind.data.frame(row = as.numeric(sapply(samDatDifVc, function(samC) which(rownames(samDF) == samC))), | |
| 205 ## name = samDatDifVc)) | |
| 206 ## } | |
| 207 | |
| 208 ## if(nrow(datMN) != nrow(samDF)) { | |
| 209 ## cat("\n\nThe dataMatrix has ", nrow(datMN), " columns (ie samples) whereas the sampleMetadata has ", nrow(samDF), " rows\n", sep="") | |
| 210 ## } else if(identical(gsub("^X", "", rownames(datMN)), rownames(samDF))) { | |
| 211 ## cat("\n\nThe dataMatrix column names start with an 'X' but not the sampleMetadata row names\n", sep="") | |
| 212 ## } else if(identical(gsub("^X", "", rownames(samDF)), rownames(datMN))) { | |
| 213 ## cat("\n\nThe sampleMetadata row names start with an 'X' but not the dataMatrix column names\n", sep="") | |
| 214 ## } else if(identical(sort(rownames(datMN)), sort(rownames(samDF)))) { | |
| 215 ## cat("\n\nThe dataMatrix column names and the sampleMetadata row names are not in the same order:\n", sep="") | |
| 216 ## print(cbind.data.frame(indice = 1:nrow(datMN), | |
| 217 ## dataMatrix_columnnames=rownames(datMN), | |
| 218 ## sampleMetadata_rownames=rownames(samDF))[rownames(datMN) != rownames(samDF), , drop = FALSE]) | |
| 219 ## } else { | |
| 220 ## cat("\n\nThe dataMatrix column names and the sampleMetadata row names are not identical:\n", sep="") | |
| 221 ## print(cbind.data.frame(indice = 1:nrow(datMN), | |
| 222 ## dataMatrix_columnnames=rownames(datMN), | |
| 223 ## sampleMetadata_rownames=rownames(samDF))[rownames(datMN) != rownames(samDF), , drop = FALSE]) | |
| 224 ## } | |
| 225 | |
| 226 ## } | |
| 227 ## datRowVc <- rownames(datMN) | |
| 228 ## datRowMakVc <- make.names(datRowVc, unique = TRUE) | |
| 229 ## if(datRowMakVc != datRowVc) { | |
| 230 ## rownames(datMN) <- datRowMakVc | |
| 231 ## cat("\n\nMessage: The sample names of the dataMatrix have been converted to the standard R format\n") | |
| 232 ## } | |
| 233 | |
| 234 ## datColVc <- colnames(datMN) | |
| 235 ## datColMakVc <- make.names(datColVc, unique = TRUE) | |
| 236 ## if(datColMakVc != datColVc) { | |
| 237 ## colnames(datMN) <- datColMakVc | |
| 238 ## cat("\n\nMessage: The variable names of the dataMatrix have been converted to the standard R format\n") | |
| 239 ## } | |
| 240 | |
| 241 ## samRowVc <- rownames(datMN) | |
| 242 ## samRowMakVc <- make.names(samRowVc, unique = TRUE) | |
| 243 ## if(samRowMakVc != samRowVc) { | |
| 244 ## rownames(datMN) <- samRowMakVc | |
| 245 ## cat("\n\nMessage: The sample names of the sampleMetadata have been converted to the standard R format\n") | |
| 246 ## } | |
| 247 | |
| 248 ## datRowVc <- rownames(datMN) | |
| 249 ## datRowMakVc <- make.names(datRowVc, unique = TRUE) | |
| 250 ## if(datRowMakVc != datRowVc) { | |
| 251 ## rownames(datMN) <- datRowMakVc | |
| 252 ## cat("\n\nMessage: The sample names of the dataMatrix have been converted to the standard R format\n") | |
| 253 ## } | |
| 254 | |
| 255 ## } | |
| 256 | |
| 257 ## checking names (optional) | |
| 258 | |
| 259 | |
| 260 | |
| 261 | |
| 262 ## datRowMakVc <- make.names(datRowVc, unique = TRUE) | |
| 263 ## if(datRowMakVc != datRowVc) { | |
| 264 ## if(makNamL) { | |
| 265 ## rownames(datMN) <- datRowMakVc | |
| 266 ## cat("\n\nMessage: The sample names of the dataMatrix have been converted to the standard R format; select the make names argument to convert them\n") | |
| 267 ## } else { | |
| 268 ## cat("\n\nWarning: Some of the sample names of the dataMatrix are not in the standard R format; select the make names argument to convert them\n") | |
| 269 ## } | |
| 270 | |
| 271 ## if(makNamL) { | |
| 272 | |
| 273 ## rownames(datMN) <- make.names(rownames(datMN), unique = TRUE) | |
| 274 ## colnames(datMN) <- make.names(colnames(datMN), unique = TRUE) | |
| 275 ## rownames(samDF) <- make.names(rownames(samDF), unique = TRUE) | |
| 276 ## rownames(varDF) <- make.names(rownames(varDF), unique = TRUE) | |
| 277 | |
| 278 ## } | |
| 279 ## checking sample names | |
| 280 | |
| 281 ## if(nrow(datMN) == nrow(samDF)) { | |
| 282 | |
| 283 | |
| 284 | |
| 285 ## } | |
| 286 | |
| 287 ## chkL <- FALSE | |
| 288 | |
| 289 ## datSamDifVc <- setdiff(rownames(datMN), rownames(samDF)) | |
| 290 | |
| 291 ## if(length(datSamDifVc)) { | |
| 292 ## cat("\nThe following samples were found in the dataMatrix column names but not in the sampleMetadata row names:\n", sep="") | |
| 293 ## print(cbind.data.frame(col = as.numeric(sapply(datSamDifVc, function(samC) which(rownames(datMN) == samC))), | |
| 294 ## name = datSamDifVc)) | |
| 295 ## } | |
| 296 | |
| 297 ## samDatDifVc <- setdiff(rownames(samDF), rownames(datMN)) | |
| 298 | |
| 299 ## if(length(samDatDifVc)) { | |
| 300 ## cat("\n\nThe following samples were found in the sampleMetadata row names but not in the dataMatrix column names:\n", sep="") | |
| 301 ## print(cbind.data.frame(row = as.numeric(sapply(samDatDifVc, function(samC) which(rownames(samDF) == samC))), | |
| 302 ## name = samDatDifVc)) | |
| 303 ## } | |
| 304 | |
| 305 ## if(nrow(datMN) != nrow(samDF)) { | |
| 306 ## cat("\n\nThe dataMatrix has ", nrow(datMN), " columns (ie samples) whereas the sampleMetadata has ", nrow(samDF), " rows\n", sep="") | |
| 307 ## } else if(identical(gsub("^X", "", rownames(datMN)), rownames(samDF))) { | |
| 308 ## cat("\n\nThe dataMatrix column names start with an 'X' but not the sampleMetadata row names\n", sep="") | |
| 309 ## } else if(identical(gsub("^X", "", rownames(samDF)), rownames(datMN))) { | |
| 310 ## cat("\n\nThe sampleMetadata row names start with an 'X' but not the dataMatrix column names\n", sep="") | |
| 311 ## } else if(identical(sort(rownames(datMN)), sort(rownames(samDF)))) { | |
| 312 ## cat("\n\nThe dataMatrix column names and the sampleMetadata row names are not in the same order:\n", sep="") | |
| 313 ## print(cbind.data.frame(indice = 1:nrow(datMN), | |
| 314 ## dataMatrix_columnnames=rownames(datMN), | |
| 315 ## sampleMetadata_rownames=rownames(samDF))[rownames(datMN) != rownames(samDF), , drop = FALSE]) | |
| 316 ## } else { | |
| 317 ## cat("\n\nThe dataMatrix column names and the sampleMetadata row names are not identical:\n", sep="") | |
| 318 ## print(cbind.data.frame(indice = 1:nrow(datMN), | |
| 319 ## dataMatrix_columnnames=rownames(datMN), | |
| 320 ## sampleMetadata_rownames=rownames(samDF))[rownames(datMN) != rownames(samDF), , drop = FALSE]) | |
| 321 ## } | |
| 322 | |
| 323 ## } | |
| 324 | |
| 325 | |
| 326 ## if(!identical(colnames(datMN), rownames(varDF))) { | |
| 327 ## ## checking variable names | |
| 328 | |
| 329 ## chkL <- FALSE | |
| 330 | |
| 331 ## datVarDifVc <- setdiff(colnames(datMN), rownames(varDF)) | |
| 332 | |
| 333 ## if(length(datVarDifVc)) { | |
| 334 ## cat("\nThe following variables were found in the dataMatrix row names but not in the variableMetadata row names:\n", sep="") | |
| 335 ## print(cbind.data.frame(row = as.numeric(sapply(datVarDifVc, function(varC) which(colnames(datMN) == varC))), | |
| 336 ## name = datVarDifVc)) | |
| 337 | |
| 338 ## } | |
| 339 | |
| 340 ## varDatDifVc <- setdiff(rownames(varDF), colnames(datMN)) | |
| 341 | |
| 342 ## if(length(varDatDifVc)) { | |
| 343 ## cat("\n\nThe following variables were found in the variableMetadata row names but not in the dataMatrix row names:\n", sep="") | |
| 344 ## print(cbind.data.frame(row = as.numeric(sapply(varDatDifVc, function(varC) which(rownames(varDF) == varC))), | |
| 345 ## name = varDatDifVc)) | |
| 346 ## } | |
| 347 | |
| 348 ## if(ncol(datMN) != nrow(varDF)) { | |
| 349 ## cat("\n\nThe dataMatrix has ", nrow(datMN), " rows (ie variables) whereas the variableMetadata has ", nrow(varDF), " rows\n", sep="") | |
| 350 ## } else if(identical(sort(colnames(datMN)), sort(rownames(varDF)))) { | |
| 351 ## cat("\n\nThe dataMatrix row names and the variableMetadata row names are not in the same order:\n", sep="") | |
| 352 ## print(cbind.data.frame(row = 1:ncol(datMN), | |
| 353 ## dataMatrix_rownames=colnames(datMN), | |
| 354 ## variableMetadata_rownames=rownames(varDF))[colnames(datMN) != rownames(varDF), , drop = FALSE]) | |
| 355 ## } else { | |
| 356 ## cat("\n\nThe dataMatrix row names and the variableMetadata row names are not identical:\n", sep="") | |
| 357 ## print(cbind.data.frame(row = 1:ncol(datMN), | |
| 358 ## dataMatrix_rownames=colnames(datMN), | |
| 359 ## variableMetadata_rownames=rownames(varDF))[colnames(datMN) != rownames(varDF), , drop = FALSE]) | |
| 360 ## } | |
| 361 ## } | |
| 362 ## checkF <- function(datInpMN, | |
| 363 ## samInpDF, | |
| 364 ## varInpDF) { | |
| 365 | |
| 366 ## mode(datInpMN) == "numeric" && | |
| 367 ## identical(rownames(datInpMN), rownames(samInpDF)) && | |
| 368 ## identical(colnames(datInpMN), rownames(varInpDF)) | |
| 369 | |
| 370 | |
| 371 ## } |