Mercurial > repos > nml > combine_assemblystats

annotate combine_stats.pl @ 0:1855203c2e6c draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
author nml
date Wed, 08 Nov 2017 16:38:50 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
1 #!/usr/bin/env perl
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
2 use strict;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
3 use warnings;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
4 use autodie;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
5 use Getopt::Long;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
6
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
7 #quick and dirty script to combine a list of assembly stats tab files into a simple csv file where each row is one strain
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
8
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
9
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
10 my ($files,$output) = prepare_inputs();
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
11
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
12
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
13 my @strains = sort { $a cmp $b } keys %{$files};
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
14
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
15
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
16 #get first file so we can determine the header
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
17 my $first = shift @strains;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
18 my $top_header;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
19 my $second_header;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
20
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
21
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
22
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
23 open my $out,'>',$output;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
24 process($first,$files->{$first},$out,1);
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
25
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
26
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
27 foreach my $name( @strains) {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
28 process($name,$files->{$name},$out);
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
29 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
30
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
31
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
32
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
33 close $out;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
34
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
35
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
36 exit;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
37
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
38 sub process {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
39 my ($name,$file,$out,$header) = @_;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
40
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
41 my @header = ("Strain");
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
42 my @values = ($name);
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
43
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
44 open my $in,'<',$file;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
45 while ( <$in>) {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
46 chomp;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
47
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
48 if (length $_ ==0) {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
49 next;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
50 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
51
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
52 #if we hit this section, we are done reading this file since the rest we do not care about
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
53 if ( $_ =~ /Simple Din.*repeats/) {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
54 last;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
55 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
56
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
57
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
58 my ($key,$value) = split /:/;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
59
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
60 #trim out the tabs
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
61 $key =~ s/\t//g;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
62 $value =~ s/\t//g;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
63
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
64 if ( $value) {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
65 push @header,$key;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
66 push @values,$value;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
67 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
68
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
69 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
70
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
71
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
72 close $in;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
73
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
74 #check to see if we are printing out the header
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
75 if ( $header) {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
76 print $out join ("\t",@header) . "\n";
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
77 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
78 print $out join ("\t",@values) . "\n";
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
79
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
80 return;
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
81 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
82
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
83
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
84 sub prepare_inputs {
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
85
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
86 my ($output,%files);
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
87
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
88
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
89
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
90 if (!GetOptions('stats=s' => \%files,
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
91 'output=s' => \$output
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
92 )){
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
93
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
94 die "Invalid options given\n";
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
95 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
96
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
97
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
98 if ( scalar keys %files == 0){
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
99 die "No files given\n";
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
100 }
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
101
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
102 return (\%files,$output);
1855203c2e6c planemo upload for repository https://github.com/phac-nml/galaxy_tools commit 132092ff7fe1c4810d1221054419389180b81657
nml
parents:
diff changeset
103 }