Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics
annotate phyloconversion/beautify_fasta.pl @ 0:5b9a38ec4a39 draft default tip
First commit of old repositories
author | osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> |
---|---|
date | Tue, 11 Mar 2014 12:19:13 -0700 |
parents | |
children |
rev | line source |
---|---|
0
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
1 #!/usr/bin/perl -w |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
2 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
3 use strict; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
4 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
5 use FindBin; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
6 use lib "$FindBin::Bin/lib"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
7 use Bio::DB::Fasta; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
8 use Bio::SeqIO; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
9 use Bio::Seq; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
10 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
11 #inputs |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
12 my $infile=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
13 my $outfile=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
14 my $delpipes=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
15 my $convgi=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
16 my $delslash=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
17 my $renumber=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
18 my $space=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
19 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
20 my $seqid; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
21 my $newnumbers=1; #for sequential renumbering of header |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
22 # open infile fasta file |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
23 my $in_obj = Bio::SeqIO->new(-file => $infile, '-format' =>'fasta'); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
24 open FILE, ">$outfile" or die $!; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
25 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
26 while (my $seq = $in_obj->next_seq() ) { |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
27 my $sequence = $seq->seq; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
28 $seqid = $seq->id; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
29 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
30 if($delslash eq 'yes'){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
31 $seqid =~ s/\\/_/g; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
32 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
33 if($convgi eq 'yes'){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
34 $seqid =~ s/gi\|/gi_/g; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
35 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
36 if($delpipes eq 'yes'){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
37 $seqid =~ s/\|/ /g; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
38 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
39 $sequence =~ s/\n//g; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
40 $sequence =~ tr/a-z/A-Z/; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
41 print FILE ">"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
42 if($renumber eq 'yes'){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
43 print FILE $newnumbers; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
44 if($space eq 'yes'){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
45 print FILE " "; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
46 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
47 $newnumbers++; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
48 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
49 print FILE $seqid." ".$seq->desc."\n".$sequence."\n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
50 } |