annotate CountUniqueIDs.pl @ 3:856008c4a5f3 draft default tip

Version 1.0.2 (updates bioperl to 1.7.2)
author pjbriggs
date Fri, 05 Oct 2018 05:33:31 -0400
parents 2f48cf393d25
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
1 #! /usr/bin/perl -w
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
2
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
3 use strict;
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
4
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
5 #### Read thru a GFF file of motifs return the number of unique ids
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
6 #### Ian Donaldson Sept 2008
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
7
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
8 #### Usage
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
9 unless(@ARGV == 2) {
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
10 die("USAGE: $0 | GFF file | Output file\n\n");
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
11 }
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
12
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
13 #### Ready output file
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
14 open(GFF, "<$ARGV[0]") or die("Could not open GFF file!!\n\n");
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
15 open(OUTPUT, ">$ARGV[1]") or die("Could not open output file!!\n\n");
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
16
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
17 #### Hash to hold ids
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
18 my %id_hash = ();
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
19
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
20 #### Work thru GFF file
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
21 while(defined(my $gff_line = <GFF>)) {
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
22 if($gff_line =~ /(^#|^\s)/) { next }
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
23
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
24 my @gff_line_bits = split(/\t/, $gff_line);
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
25
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
26 my $id = $gff_line_bits[0];
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
27
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
28 $id_hash{$id}=1;
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
29 }
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
30
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
31 my @all_keys = sort(keys(%id_hash));
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
32
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
33 my $elements = scalar(@all_keys);
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
34
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
35 #print OUTPUT "There are $elements unique sequences in the file\n";
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
36 print OUTPUT "$elements non-redundant sequences\n";
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
37
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
38 #### Close files
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
39 close(GFF);
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
40 close(OUTPUT);
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
41
2f48cf393d25 Add Perl scripts missing from previous upload.
pjbriggs
parents:
diff changeset
42 exit;