scripts/biosql/terms/interpro2go.pl

   1 #!/usr/bin/perl
   2
   3 use strict;
   4
   5 while (<>) {
   6         # Process STDIN (the interpro2go file from geneontology.org
   7         # For each line, output to STDOUT the following:
   8         #       sourceaccession (IPR),targetaccession (GO)
   9         # where dbid is the human-readable non-unique IPR reference,
  10         # dbaccession is the unique IPR reference, and goaccession is the
  11         # unique GO accession.
  12
  13         # Interpro lines look like this:
  14         # InterPro:IPR000018 P2Y4 purinoceptor > GO:purinergic nucleotide receptor activity, G-protein coupled ; GO:0045028
  15
  16         /^InterPro:/ or next;
  17         s/^InterPro:(IPR\d+).*>.*;\s+(GO:\d+)/$1,$2/;
  18         print;
  19 }