Advertisement
ProzacR

another text parser

Feb 22nd, 2013
404
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
  1. #!/usr/bin/perl
  2. #
  3. # nuskaityti bindingdb sdf failus
  4. # H-Se/DB26984_50278320.sdf kaip testas imtas
  5. #
  6. # VR
  7. #
  8.  
  9. #use warnings;
  10. use Data::Dumper;
  11.  
  12.  
  13. #nuskaityti failus i @molekules
  14. die "usage VR_bindingdb_parse.pl file.sdf" if (@ARGV != 1);
  15. $FILE = $ARGV[0];
  16. open FILE, "<", $FILE or die $!;
  17. $x=0;
  18. $key='str';
  19. while(<FILE>) {
  20.  chop($_);
  21.  next if ($_ =~ /^$/);   #jei eilute tuscia praleidziam
  22.  if ($_ =~ /\$\$\$\$/) { #jei eiluteje yra: $$$$
  23.   $x++;
  24.   $key = 'str';
  25.   next;
  26.  }
  27.  #print $x, $_;
  28.  if ($_ =~ /^\>/) {      #jei eilute prasideda: >
  29.   chop();
  30.   $_ =~ s/>  <//;
  31.   $key = $_;
  32.  } else {
  33.   #$molekules[molekules nr.]{irasas}
  34.   push @{ $molekules[$x]{$key} }, $_;
  35.  }
  36. }
  37. close FILE;
  38. #print STDERR Dumper \@molekules;
  39. print $FILE, "\n";
  40.  
  41.  
  42. #rasti nuo kur prasideda kitokia molekule
  43. #pirmoji kita yra $ta_pati+1
  44. kita();
  45. sub kita {
  46. $y=0;
  47. print $y;
  48. $ta_pati=0;
  49. while($molekules[$y]{'str'}) {
  50.  #print STDERR Dumper $molekules[$y]{'str'}, "\n";
  51.  $z=0;
  52.  while($molekules[$y]{'str'}[$z]) {
  53.   #jei z eilute nesutampa y ir y+1 nors karta tai kita molekule
  54.   if ($molekules[$y]{'str'}[$z] ne $molekules[$y+1]{'str'}[$z]) {
  55.    $ta_pati=$y;
  56.    #print STDERR $y, " :", $molekules[$y]{'str'}[$z], " ir ", $molekules[$y+1]{'str'}[$z], "\n";
  57.    last;
  58.   }
  59.   $z++;
  60.  }
  61.  last if ($ta_pati);
  62.  $y++;
  63. }
  64. }
  65. #print STDERR Dumper $molekules[0], "\n";
  66.  
  67.  
  68. #Surinkti Ki vertes i du array:
  69. $x=0;
  70. while($molekules[$x]{'Enzymologic: Ki nM'}) {
  71. #tik jei Ki apibrezta
  72. if ($molekules[$x]{'Enzymologic: Ki nM'} > 0) {
  73.  if ($x <= $ta_pati) {
  74.   push @pirma, $molekules[$x]{'Enzymologic: Ki nM'};
  75.  } else {
  76.   push @antra, $molekules[$x]{'Enzymologic: Ki nM'};
  77.  }
  78. }
  79. $x++;
  80. }
  81. print STDERR "Ki ", $molekules[0]{'BindingDB Display Name'}[0], " molekulei:\n";
  82. print STDERR Dumper @pirma;
  83. print STDERR "Ki ", $molekules[$ta_pati+1]{'BindingDB Display Name'}[0], " molekulei:\n";
  84. print STDERR Dumper @antra;
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement