tools/annot/annotmix.pl

   1 #!/usr/bin/perl -w
   2 #use threads 1.73;
   3 use strict;
   4 use warnings;
   5 use Time::HiRes qw ( gettimeofday tv_interval );
   6 use Galaxy::ShowHelp;
   7 #use Galaxy::Data;
   8 #use Fcntl qw(:DEFAULT :flock);
   9
  10 $main::VERSION=0.1.1;
  11
  12 our $opts='i:o:bv';
  13 our($opt_i, $opt_o, $opt_v,$opt_b);
  14
  15 our $help=<<EOH;
  16 \t-i Annotation files list (./anno.lst) [ID\\tPath_to_file\\n]
  17 \t-o Output mixed file (mixed_annot.txt)
  18 \t-v show verbose info to STDOUT
  19 \t-b No pause for batch runs
  20 EOH
  21 # \t-d dbSNP SQLite data file (_tdbSNP.sqlite)
  22
  23 ShowHelp();
  24
  25 $opt_i='./anno.lst' if ! defined $opt_i;
  26 $opt_o='mixed_annot.txt' if ! $opt_o;
  27
  28 print STDERR "From [$opt_i] to [$opt_o]\n";
  29 if (! $opt_b) {print STDERR 'press [Enter] to continue...'; <>;}
  30
  31 my $start_time = [gettimeofday];
  32
  33 my (@Annots,%Keys);
  34 open( SAMP,'<',$opt_i) or die "Error: $!\n";
  35 open( OUT,'>',$opt_o) or die "Error: $!\n";
  36 print OUT 'Gene';
  37 while (<SAMP>) {
  38         chomp;
  39         my ($id,$file)=split /\s+/;
  40         my %dat;
  41         getHash($file,\%dat);
  42         ++$Keys{$_} for keys %dat;
  43         push @Annots,[$id,\%dat];
  44         print OUT "\t$id\t${id}_nfo";
  45 }
  46 close SAMP;
  47 print OUT "\n";
  48
  49 sub getHash {
  50         my( $file ,$hash) = @_;
  51         open (F,"$file") or die $!;
  52         while (<F>) {
  53                 chomp;
  54                 my($key,$a,$b)=split /\s+/,$_,3;
  55                 $key=~s/\.\d+$//;
  56                 $a=~s/\t/ /g;
  57                 $b=~s/\t/ /g;
  58                 my $value = "$a\t$b";
  59         #print "$key\t$value\n";
  60                 $$hash{$key} = $value;
  61         }
  62         close F;
  63 }
  64
  65
  66 for my $gene (sort keys %Keys) {
  67         print OUT $gene;
  68         for (@Annots) {
  69                 my ($id,$dathash)=@$_;
  70                 my $v=(defined $$dathash{$gene})?$$dathash{$gene}:".\t.";
  71                 print OUT "\t$v";
  72         }
  73         print OUT "\n";
  74 }
  75
  76 close OUT;
  77
  78 my $stop_time = [gettimeofday];
  79 #$|=1;
  80 print STDERR "\n Time Elapsed:\t",tv_interval( $start_time, $stop_time )," second(s)\n";
  81
  82 print STDERR "\033[32;1mgrep -i 'transcription factor' mixed_annot.txt > out\033[0;0m\n";
  83
  84 __END__