#!/usr/local/bin/perl
use strict;
use warnings;
open (FH,'<',"F:/Bioinformatics_NCBI/20MARCH_10/PERL Analysis/test.txt") or die("$!\n");
open(FO, '>',"F:/Bioinformatics_NCBI/20MARCH_10/PERL Analysis/testOut.txt") or die ("$!\n");   #TESTING
my (@snoRNA, @geneID, @productID, @geneNames, @references,@queries,@subjects);
while(<FH>){
        chomp;
        if(/(?=^\d+$)/../(?=http:.*)\n/){ #range matching
               # s/\W+\n+!\W+//;
               next unless /(\w+ |\| | \n+)/x;  #except for words | pipes | \n
                print FO $_, "\n" ;
        }
        if(/snoRNA(\s+|\d+)[\s|-|\d]/){   #snoRNA
        push @snoRNA, $_;
                }
         if(/^\d+$/){      #exon Numbers
                push @exonNumbers, $_;
                }
                if(/^GI:\d+[\.\d+]/){    #gene Names
                push @geneID , $_;
                }
        if(/^NM_\d+[\.\d+]/){        #gene product ID
                my $name = $_;
                $name =~ s/\s+$//; #substitute the trailing blanks..
                push @productID, $name;
                }
        if(/homo sapiens[\w+\W+]/i){ #gene name, Need MultiLine support..
                my $name = $_;
                push @geneNames, $name;
                }
        if(/http:.*/){          #web refs, need multiline support..
                my $name = $_;
                push @references, $name;
                }
        if(/^Query(\s+)\d+\s+[agtc]/i){       #Prepare the query and subject arrays
                my $queryName = $_;
                $queryName =~ s/$1//;
                push @queries, $queryName;
                }
        if(/^sbjct(\s+)\d+\s+[agtc]/i){
                        my $sbjctName =  $_;
                        $sbjctName =~ s/$1//;
                        push @subjects, $sbjctName;
                        }

        }