fasta-shuffle-letters

来自「EM算法的改进」· 代码 · 共 102 行

#!/usr/bin/perl
#
# $Id: fasta-shuffle-letters.txt 1339 2006-09-21 19:46:28Z tbailey $
# $Log$
# Revision 1.1  2005/07/28 23:54:37  nadya
# Initial revision
#
#
# AUTHOR: Timothy L. Bailey
# CREATE DATE: 12/4/98
 
$pgm = $0;                      # name of program
$pgm =~ s#.*/##;                # remove part up to last slash
@args = @ARGV;                  # arguments to program
$| = 1;                         # flush after all prints
$SIG{'INT'} = 'cleanup';        # interrupt handler
# Note: so that interrupts work, always use for system calls:
#       if ($status = system($command)) {&cleanup($status)}
 
# requires
push(@INC, split(":", $ENV{'PATH'}));   # look in entire path
 
# defaults
 
$usage = <<USAGE;               # usage message
  USAGE:
        $pgm

	-tod		use time of day as seed for the random number generator
			default: 0

	Shuffle each sequence in a file of fasta sequences.
	(Does not shuffle the order of the sequences.)

	Reads file from standard input.
	Writes to standard output.
USAGE

$nargs = 0;                     # number of required args
if ($#ARGV+1 < $nargs) { die($usage); }
 
# get input arguments
while ($#ARGV >= 0) {
  $_ = shift;
  if ($_ eq "-h") {
    die($usage);
  } elsif ($_ eq "-tod") {
    $tod = 1;
  } else {
    die($usage);
  }
}

#
# intialize the random number generator if -tod 
#
if ($tod) {
  srand(); 
} else {
  srand(0); 
}

while (<STDIN>) {
  chop;				# remove newline
  if (/^>/) {			# new sequence
    if ($label) { print &shuffle($label, $seq); }
    $label = $_;		# save label
    $seq = "";			# initialize sequence
  } else {			# sequence
    $seq .= $_;			# save line of sequence
  }
}
# shuffle and print last sequence
print &shuffle($label, $seq);

################################################################################
#                       Subroutines                                            #
################################################################################
sub shuffle {
  my($label, $seq) = @_;
  my($out, @chars, $n, $i, %hash, @keys, $out, $x);

  $out = $label;					# put label
  @chars = split(//, $seq);				# split into characters
  $n = @chars;

  # Give each position in the sequence a random key and sort the keys
  for ($i=0; $i<$n; $i++) {
    do {$x = rand()/(1+rand());} while (defined $hash{"$x"});
    $x = rand()/(1+rand());				# only 32768 values for rand!
    $hash{"$x"} = $chars[$i];
  }
  @keys = sort {$a <=> $b} keys %hash;

  for ($i=0; $i<$n; $i++) {				# shuffle sequence
    $out .= "\n" if ($i%70 == 0);			# split at 70 columns
    $out .= $hash{$keys[$i]};				# add to shuffled seq
  }
  undef %hash;
  $out .= "\n";
}