perl/etc/WoodyMiaoLin/PbeBefore2015/BankIt_submit.pl

   1 #!/usr/bin/perl
   2 use strict;
   3 use warnings;
   4
   5 open I1, "<", "../1.find_need_to_submit/MT/MT_needto.fasta";
   6 open I2, "<", "../1.find_need_to_submit/Y/Y_needto.fasta";
   7 open I3, "<", "../1.find_need_to_submit/PLP/PLP_needto.fasta";
   8 #open O, ">", "BankIt_submit.fasta";
   9
  10 my %STS;
  11 while (<I1>) {
  12         /^>(\w{3})(MT\d+)/;
  13         my $s = <I1>;
  14         $STS{"16S"}{$1}{$2} = substr $s, 0, 367, "";
  15         $STS{ATP8}{$1}{$2} = substr $s, 0, 176, "";
  16         $STS{CytB}{$1}{$2} = substr $s, 0, 1249, "";
  17         warn $s unless $s eq "\r\n";
  18         delete $STS{ATP8};
  19 }
  20 close I1;
  21 while (<I2>) {
  22         /^>(\w{3})(Y)/;
  23         my $s = <I2>;
  24         $STS{SMCY3}{$1}{$2} = substr $s, 0, 841, "";
  25         $STS{SMCY7_STR_upstream}{$1}{$2} = substr $s, 0, 224, "";
  26         $STS{SMCY7_STR_downstream}{$1}{$2} = substr $s, 0, 325, "";
  27         $STS{DBY7}{$1}{$2} = substr $s, 0, 280, "";
  28         $STS{UTY11}{$1}{$2} = substr $s, 0, 484, "";
  29         warn $s unless $s eq "\r\n";
  30 }
  31 close I2;
  32 while (<I3>) {
  33         $/ = ">";
  34         my $s = <I3>;
  35         chomp $s;
  36         $s =~ s/\s//g;
  37         $/ = "\n";
  38         /^>?(\w{3})([\--Z]+)/;
  39         $STS{PLP}{$1}{$2} = $s;
  40 }
  41 close I3;
  42
  43 my %SOURCE = (
  44         "Pbe"   =>      "Prionailurus bengalensis",
  45         "Pvi"   =>      "Prionailurus viverrinus",
  46         "Ipl"   =>      "Prionailurus planiceps",
  47         "Pte"   =>      "Pardofelis temminckii",
  48         "Pma"   =>      "Pardofelis marmorata",
  49         "Pti"   =>      "Panthera tigris",
  50         "Ppa"   =>      "Panthera pardus",
  51 );
  52
  53 my %desc = (
  54         PLP => 'X chromosome, PLP1, partial sequence',
  55         CytB => 'mitochondrial, cytochrome b complete CDS, tRNA-Thr complete sequence, tRNA-Pro partial sequence',
  56         '16S' => 'mitochondrial, 16S ribosomal RNA, partial sequence',
  57         UTY11 => 'Y chromosome, UTY, partial sequence',
  58         DBY7 => 'Y chromosome, DBY, partial sequence',
  59         SMCY3 => 'Y chromosome, SMCY, partial sequence',
  60         SMCY7_STR_upstream => 'Y chromosome, SMCY, partial sequence',
  61         SMCY7_STR_downstream => 'Y chromosome, SMCY, partial sequence'
  62 );
  63
  64 foreach my $a (sort keys %STS) {
  65         open O, ">", "submit_$a.fa";
  66         foreach my $b (sort keys %{$STS{$a}}) {
  67                 foreach my $c (sort keys %{$STS{$a}{$b}}) {
  68                         $STS{$a}{$b}{$c} =~ s/-|\?//g;
  69                         my $l = length $STS{$a}{$b}{$c};
  70                         next if $l < 200;
  71                         next unless $SOURCE{$b};
  72                         my $str = "$SOURCE{$b}, $desc{$a}";
  73                         if ($a eq "PLP") {
  74                                 print O ">$b$c [organism=$SOURCE{$b}] [chromosome=X] [gcode=1] $str\n$STS{$a}{$b}{$c}\n";
  75                         } elsif ($a eq "CytB") {
  76                                 if ($l > 1000) {
  77                                         print O ">$a-$b$c [organism=$SOURCE{$b}] [location=mitochondrion] [mgcode=2] $str\n$STS{$a}{$b}{$c}\n";
  78                                 } else {
  79                                         print O ">$a-$b$c [organism=$SOURCE{$b}] [location=mitochondrion] [mgcode=2]";
  80                                         print O "$SOURCE{$b}, mitochondrial, cytochrome b, partical CDS\n$STS{$a}{$b}{$c}\n";
  81                                 }
  82                         } elsif ($a eq "16S"){
  83                                 print O ">$a-$b$c [organism=$SOURCE{$b}] [location=mitochondrion] $str\n$STS{$a}{$b}{$c}\n";
  84                         } elsif ($a =~ /^(SMCY|DBY7|UTY11)/) {
  85                                 print O ">$a-$b$c [organism=$SOURCE{$b}] [chromosome=Y] [gcode=1] $str\n$STS{$a}{$b}{$c}\n";
  86                         } else {
  87                                 print O ">$a-$b$c [organism=$SOURCE{$b}] [gcode=1]\n$STS{$a}{$b}{$c}\n";
  88                                 warn "Bad gene name!";
  89                         }
  90                 }
  91         }
  92         close O;
  93 }
  94 #close O;