Mercurial > repos > iuc > fasta_stats

annotate fasta-stats.pl @ 0:9c620a950d3a draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
author iuc
date Thu, 22 Nov 2018 04:16:35 -0500
parents
children 16f1f3e2de42
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
1 #!/usr/bin/env perl
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
2
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
3 # fasta-stats
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
4 # written by torsten.seemann@monash.edu
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
5 # oct 2012
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
6
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
7 use strict;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
8 use warnings;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
9 use List::Util qw(sum min max);
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
10
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
11 # stat storage
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
12
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
13 my $n=0;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
14 my $seq = '';
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
15 my %stat;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
16 my @len;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
17
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
18 # MAIN LOOP collecting sequences
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
19
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
20 while (my $line = <ARGV>) {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
21 chomp $line;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
22 if ($line =~ m/^\s*>/) {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
23 process($seq) if $n;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
24 $n++;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
25 $seq='';
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
26 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
27 else {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
28 $seq .= $line;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
29 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
30 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
31
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
32 process($seq) if $n;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
33
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
34 # sort length array
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
35 # (should use hash here for efficiency with huge no of short reads?)
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
36
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
37 @len = sort { $a <=> $b } @len;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
38
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
39 # compute more stats
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
40
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
41 $stat{'num_seq'} = scalar(@len);
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
42
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
43 if (@len) {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
44 $stat{'num_bp'} = sum(@len);
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
45 $stat{'len_min'} = $len[0];
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
46 $stat{'len_max'} = $len[-1];
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
47 $stat{'len_median'} = $len[int(@len/2)];
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
48 $stat{'len_mean'} = int( $stat{'num_bp'} / $stat{'num_seq'} );
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
49 # calculate n50
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
50
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
51 $stat{'len_N50'} = 0;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
52 my $cum=0;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
53 my $thresh = int 0.5 * $stat{'num_bp'};
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
54 for my $i (0 .. $#len) {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
55 $cum += $len[$i];
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
56 if ($cum >= $thresh) {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
57 $stat{'len_N50'} = $len[$i];
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
58 last;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
59 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
60 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
61 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
62
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
63 #calculate GC content
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
64 $stat{'num_bp_not_N'} = $stat{'num_G'} + $stat{'num_C'} + $stat{'num_A'} + $stat{'num_T'};
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
65 $stat{'GC_content'} = ($stat{'num_G'} + $stat{'num_C'}) / $stat{'num_bp_not_N'}*100;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
66
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
67 # print stats as .tsv
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
68
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
69 for my $name (sort keys %stat) {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
70 if ($name =~ m/GC_content/){
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
71 printf "%s\t%0.1f\n", $name, $stat{$name};
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
72 } else {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
73 printf "%s\t%s\n", $name, $stat{$name};
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
74 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
75 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
76
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
77 # run for each sequence
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
78
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
79 sub process {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
80 my($s) = @_;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
81 # base composition
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
82 for my $x (qw(A G T C N)) {
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
83 my $count = $s =~ s/$x/$x/gi;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
84 $stat{"num_$x"} += $count;
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
85 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
86 # keep list of all lengths encountered
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
87 push @len, length($s);
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
88 }
9c620a950d3a planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit d6a78405947a91659a4168ddb2f1534327f044cb
iuc
parents:
diff changeset
89