perl/etc/denovo/gethist.pl

   1 #!/bin/env perl
   2 use strict;
   3 use warnings;
   4
   5 die "Usage: $0 <col> <input[gzipped]> <output prefix>\n" if @ARGV < 3;
   6 my ($col,$inf,$outf)=@ARGV;
   7 warn "From Col=[$col] of [$inf] to [$outf]\n";
   8
   9 --$col;
  10 my ($Sum,$Count,%Cnt)=(0,0);
  11
  12 sub openFH($) {
  13         my $inf = $_[0];
  14         my $FH;
  15         if ($inf =~ /\.gz$/i) {
  16                 open $FH,'-|',"gzip -dc $inf" or die "Error opening $inf: $!\n";
  17         } elsif ($inf =~ /\.xz$/i) {
  18                 open $FH,'-|',"xz -dc $inf" or die "Error opening $inf: $!\n";
  19         } elsif ($inf =~ /\.bz2$/i) {
  20                 open $FH,'-|',"bzip2 -dc $inf" or die "Error opening $inf: $!\n";
  21         } else {
  22                 open $FH,'<',$inf or die "Error opening $inf: $!\n";
  23         }
  24         return $FH;
  25 }
  26
  27 my $FH = openFH($inf);
  28 while (<$FH>) {
  29         my @Dat = split /\s+/;
  30         next if /^[#;]/;
  31         ++$Cnt{$Dat[$col]};
  32         ++$Count;
  33         $Sum += $Dat[$col];
  34 }
  35 close $FH;
  36
  37 open O,'>',"$outf.dat" or die $!;
  38 print O "# TotalCount=$Count, ValueSum=$Sum, Average=",$Sum/$Count,"\n#Value\tCount\tCountRatio\tCumCountRatio\n";
  39
  40 my $Cum=0;
  41 for my $k (sort { $a<=> $b } keys %Cnt) { # $Cnt{$a} <=> $Cnt{$b} || $a<=> $b
  42         $Cum += $Cnt{$k}/$Count;
  43         print O join("\t",$k,$Cnt{$k},$Cnt{$k}/$Count,$Cum),"\n";
  44 }
  45 close O;
  46
  47 __END__
  48 set terminal png notransparent nointerlace size 1200,960 font '/opt/arial.ttf' 24
  49 set output "plot.png"
  50 set xlabel 'Depth'
  51 set ylabel 'Count'
  52 set xrange[2:862]
  53 set yrange[0:320000]
  54 plot 'tw.dat' using 1:2 with lines