perl/etc/WoodyMiaoLin/PbeBefore2015/collect_genetic_information.pl

   1 #!/usr/bin/perl
   2 use strict;
   3 use warnings;
   4
   5 open I0, "<", "../a.information/Haplotype_Region.txt";
   6 my %hap_reg;
   7 while (<I0>) {
   8         chomp;
   9         my @c = split /\t/;
  10         next if !$c[0];
  11         $hap_reg{$c[0]} = $c[1];
  12 }
  13 close I0;
  14
  15 open I1, "<", "../a.information/Sample_MT_X_Y.txt";
  16 my %sample_gt; # Sample, MT, X, Y, ATP8, Autosome.
  17 my %sample_re; # Sample, MT, X, Y, ATP8, Autosome.
  18 while (<I1>) {
  19         chomp;
  20         my @c = split /\t/;
  21         my $d = shift @c;
  22         $sample_gt{$d} = [@c];
  23         $sample_re{$d} = [$hap_reg{$c[0]}, $hap_reg{$c[1]}, $hap_reg{$c[2]}];
  24 }
  25 close I1;
  26
  27 open I2, "<", "/share/users/miaolin/5.Leopard_cat/3.PBE_paper_2015/g.ATP8_haplotype_tree/b.modified_alignment/f.modified_alignment/ATP8_haplotype_location_sampleID.txt";
  28 while (<I2>) {
  29         chomp;
  30         next if !/^Pbe/;
  31         my @c = split /\t/;
  32         my @d = split / /, $c[2];
  33         foreach (@d) {
  34                 push @{$sample_gt{$_}}, $c[0];
  35                 push @{$sample_re{$_}}, $hap_reg{$c[0]};
  36         }
  37 }
  38 close I2;
  39 foreach (keys %sample_gt) {
  40         if (@{$sample_gt{$_}} == 3) {
  41                 push @{$sample_gt{$_}}, 0;
  42         } elsif (@{$sample_gt{$_}} == 4) {
  43                 next;
  44         } else {
  45                 warn @{$sample_gt{$_}};
  46         }
  47 }
  48
  49 foreach (keys %sample_re) {
  50         if (@{$sample_re{$_}} == 3) {
  51                 push @{$sample_re{$_}}, 0;
  52         } elsif (@{$sample_re{$_}} == 4) {
  53                 next;
  54         } else {
  55                 warn @{$sample_re{$_}};
  56         }
  57 }
  58
  59 open I3, "<", "../a.information/Sample_AutoRegion.txt";
  60 while (<I3>) {
  61         chomp;
  62         my @c = split /\t/;
  63         push @{$sample_gt{$c[0]}}, $c[1];
  64         push @{$sample_re{$c[0]}}, $c[2];
  65 }
  66 close I3;
  67 foreach (keys %sample_gt) {
  68         if (@{$sample_gt{$_}} == 4) {
  69                 push @{$sample_gt{$_}}, 0;
  70         } elsif (@{$sample_gt{$_}} == 5) {
  71                 next;
  72         } else {
  73                 warn @{$sample_gt{$_}};
  74         }
  75 }
  76
  77 foreach (keys %sample_re) {
  78         if (@{$sample_re{$_}} == 4) {
  79                 push @{$sample_re{$_}}, 0;
  80         } elsif (@{$sample_re{$_}} == 5) {
  81                 next;
  82         } else {
  83                 warn @{$sample_re{$_}};
  84         }
  85 }
  86
  87 foreach (sort keys %sample_gt) {
  88         foreach (@{$sample_gt{$_}}) {
  89                 $_ = "n/a" unless $_;
  90         }
  91         my %region;
  92         foreach (@{$sample_re{$_}}) {
  93                 $_ = "n/a" unless $_;
  94                 $region{$_} = 1;
  95         }
  96         delete $region{"n/a"};
  97         if ((keys %region) == 1) {
  98                 push @{$sample_re{$_}}, (keys %region);
  99         } elsif ((keys %region) == 2) {
 100                 if ($region{Northern}) {
 101                         if ($region{Iriomote}) {
 102                                 push @{$sample_re{$_}}, "Iriomote";
 103                         } elsif ($region{"China/Indochina"}) {
 104                                 push @{$sample_re{$_}}, "China/Indochina";
 105                         } elsif ($region{Taiwan}) {
 106                                 push @{$sample_re{$_}}, "Taiwan";
 107                         } elsif ($region{Amur}) {
 108                                 push @{$sample_re{$_}}, "Amur";
 109                         } elsif ($region{Sunda}) {
 110                                 push @{$sample_re{$_}}, "Hybrid";
 111                         } elsif ($region{Philippines}) {
 112                                 push @{$sample_re{$_}}, "Hybrid";
 113                         } elsif ($region{Southern}) {
 114                                 push @{$sample_re{$_}}, "Hybrid";
 115                         } else {
 116                                 push @{$sample_re{$_}}, "CHECK!";
 117                         }
 118                 } elsif ($region{Southern}) {
 119                         if ($region{Iriomote}) {
 120                                 push @{$sample_re{$_}}, "Hybrid";
 121                         } elsif ($region{"China/Indochina"}) {
 122                                 push @{$sample_re{$_}}, "Hybrid";
 123                         } elsif ($region{Taiwan}) {
 124                                 push @{$sample_re{$_}}, "Hybrid";
 125                         } elsif ($region{Amur}) {
 126                                 push @{$sample_re{$_}}, "Hybrid";
 127                         } elsif ($region{Sunda}) {
 128                                 push @{$sample_re{$_}}, "Sunda";
 129                         } elsif ($region{Philippines}) {
 130                                 push @{$sample_re{$_}}, "Philippines";
 131                         } else {
 132                                 push @{$sample_re{$_}}, "CHECK!";
 133                         }
 134                 }
 135         } else {
 136                 if ($region{Northern} and $region{Southern}) {
 137                         push @{$sample_re{$_}}, "Hybrid";
 138                 } else {
 139                         push @{$sample_re{$_}}, "CHECK!";
 140                 }
 141         }
 142 }
 143
 144 open O, ">", "sample_genotype_region.txt";
 145 print O "#SampleID\tgtMT\tgtX\tgtY\tgtATP8\tqAutosome(K=2)\tregionMT\tregionX\tregionY\tregionATP8\tregionAutosome\tGeneticRegion\n";
 146 foreach (sort keys %sample_gt) {
 147         print O $_, "\t";
 148         print O join("\t", @{$sample_gt{$_}}), "\t";
 149         print O join("\t", @{$sample_re{$_}}), "\n";
 150 }
 151 close O;