Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics
annotate phyloconversion/length_outliers.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 Bio::DB::Fasta; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
6 use Bio::SeqIO; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
7 use Bio::Seq; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
8 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
9 #inputs |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
10 my $infile=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
11 my $outfile=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
12 my $deloutfile=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
13 my $percent=shift(@ARGV); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
14 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
15 my $seqid; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
16 my $newnumbers=1; #for sequential renumbering of header |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
17 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
18 open FILE, ">$outfile" or die $!; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
19 open DELFILE, ">$deloutfile" or die $!; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
20 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
21 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
22 # open infile fasta file to get average length |
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 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
25 my $seqcount; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
26 my $seqsum = 0; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
27 my $avelen; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
28 while (my $seq = $in_obj->next_seq() ) { |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
29 my $sequence = $seq->seq; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
30 my $seqlen = length($sequence); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
31 $seqcount++; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
32 $seqsum = $seqsum + $seqlen; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
33 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
34 $avelen = $seqsum/$seqcount; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
35 print "AVE= $avelen \n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
36 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
37 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
38 # open infile fasta file to get average length |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
39 $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
|
40 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
41 while (my $seq = $in_obj->next_seq() ) { |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
42 my $sequence = $seq->seq; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
43 $seqid = $seq->id; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
44 $sequence =~ s/\n//g; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
45 $sequence =~ tr/a-z/A-Z/; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
46 my $seqlen = length($sequence); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
47 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
48 if($seqlen > ($avelen * ($percent/100) ) ){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
49 print FILE ">"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
50 print FILE $seqid." ".$seq->desc."\n".$sequence."\n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
51 }else{ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
52 print "Writing sequence of $seqlen to DELFILE\n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
53 print DELFILE ">"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
54 print DELFILE $seqid." ".$seq->desc."\n".$sequence."\n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
55 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
56 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
57 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
58 close FILE; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
59 close DELFILE; |