Mercurial > repos > iss > eurl_vtec_wgs_pt
annotate scripts/MultifastaFromBlast.pl @ 5:fd9093b9368f draft
planemo upload commit 9af296d380b81335353ebfb8bb6de831e1fbebe1
author | iss |
---|---|
date | Thu, 19 Oct 2023 20:27:22 +0000 |
parents | c6bab5103a14 |
children |
rev | line source |
---|---|
0
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
1 #!/usr/bin/env perl |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
2 use strict; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
3 use warnings; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
4 use English; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
5 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
6 # Parse arguments |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
7 my ($inputs, $output) = @ARGV; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
8 # Run program |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
9 unlink $output; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
10 my @infiles = split( /,/, $inputs ); |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
11 foreach(@infiles) { |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
12 transformFileContent($_, $output); |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
13 } |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
14 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
15 # read input file and write multifasta with sequence (forward or reverse complement) |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
16 sub transformFileContent { |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
17 my ($infile, $outfile) = @_; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
18 open my $if, '<', $infile or die "Cannot open : $infile!"; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
19 open my $of, '>>', $outfile or die "Cannot open : $outfile!"; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
20 my @lines = <$if>; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
21 close $if; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
22 foreach(@lines) { |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
23 my @elems = split( /\t/, $_ ); |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
24 print $of ">$elems[0]\n"; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
25 chomp $elems[2]; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
26 if ($elems[1] == 1) { |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
27 print $of "$elems[2]\n"; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
28 } |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
29 else { |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
30 my $revcomp = reverseComplement($elems[2]); |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
31 print $of "$revcomp\n"; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
32 } |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
33 } |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
34 close $of; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
35 return 0; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
36 } |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
37 |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
38 # calculate reverse complement |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
39 sub reverseComplement { |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
40 my ($DNA) = @_; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
41 my $revcom = reverse $DNA; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
42 $revcom =~ tr/ACGTacgt/TGCAtgca/; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
43 return $revcom; |
c6bab5103a14
"planemo upload commit 6abf3e299d82d07e6c3cf8642bdea80e96df64c3-dirty"
iss
parents:
diff
changeset
|
44 } |