Commit 9f332056e88eedaa57bcc71267eb489bd80226c2
1 parent
b344449ca4
Exists in
master
Fixed annoying bug where the origial input file(s) to getDomainTopology.pl was(were) modified
Showing 1 changed file with 11 additions and 3 deletions Side-by-side Diff
getDomainTopology.pl
View file @
9f33205
| ... | ... | @@ -234,6 +234,12 @@ |
| 234 | 234 | my $tcdbDir = "$rootDir/tcdb"; |
| 235 | 235 | system "mkdir -p $tcdbDir" unless (-d $tcdbDir); |
| 236 | 236 | |
| 237 | + #to prevent modifying the original files, here I'll save the input | |
| 238 | + #sequences with the artificial TCIDs | |
| 239 | + my $origInfilesDir = "$rootDir/inputFiles"; | |
| 240 | + system "mkdir -p $origInfilesDir" unless (-d $origInfilesDir); | |
| 241 | + | |
| 242 | + | |
| 237 | 243 | #generate an empty "TCDB sequence file" that will contains proteins not in TCDB |
| 238 | 244 | my $new_tcdbSeqsFile = "$tcdbDir/tcdb.faa"; |
| 239 | 245 | system "cat /dev/null > $new_tcdbSeqsFile"; |
| ... | ... | @@ -247,6 +253,8 @@ |
| 247 | 253 | my $tcid = $pair->[0]; |
| 248 | 254 | my $tgtF = $pair->[1]; |
| 249 | 255 | |
| 256 | + my @comp = split(/\//, $tgtF); | |
| 257 | + my $tgtFileName = $comp[-1]; | |
| 250 | 258 | |
| 251 | 259 | #Add family to the main array (as if provided by the -f commandline option) |
| 252 | 260 | push (@fams, $tcid); |
| ... | ... | @@ -270,8 +278,8 @@ |
| 270 | 278 | |
| 271 | 279 | |
| 272 | 280 | #Replace the TCID in the file corresponding to the target proteins |
| 273 | - my $cmd1 = qq(perl -i.orig -pe 's/\\>([a-zA-Z0-9_-]+).*/\\>${tgtTC}-\$1/;' $tgtF); | |
| 274 | - system $cmd1 unless (-f "${tgtF}.orig"); | |
| 281 | + my $cmd1 = qq(perl -pe 's/\\>([a-zA-Z0-9_-]+).*/\\>${tgtTC}-\$1/;' $tgtF > $origInfilesDir/$tgtFileName); | |
| 282 | + system $cmd1 unless (-f "$origInfilesDir/$tgtFileName"); | |
| 275 | 283 | |
| 276 | 284 | |
| 277 | 285 | #Extract sequences for reference family |
| ... | ... | @@ -282,7 +290,7 @@ |
| 282 | 290 | |
| 283 | 291 | |
| 284 | 292 | #Add family and target sequences to the new TCDB family |
| 285 | - my $cmd3 = qq(cat $outFile $tgtF >> $new_tcdbSeqsFile); | |
| 293 | + my $cmd3 = qq(cat $outFile $origInfilesDir/$tgtFileName >> $new_tcdbSeqsFile); | |
| 286 | 294 | system $cmd3; |
| 287 | 295 | } |
| 288 | 296 |