| 1 |
gpertea |
23 |
#!/usr/bin/perl |
| 2 |
|
|
use strict; |
| 3 |
|
|
use Getopt::Std; |
| 4 |
|
|
use FindBin;use lib $FindBin::Bin; |
| 5 |
|
|
|
| 6 |
|
|
my $usage = q/Usage: |
| 7 |
|
|
dust.pl [-c <repeat#>] <fasta_file> |
| 8 |
|
|
|
| 9 |
|
|
Masks all repeats of unit length 1 or greater that are repeated at |
| 10 |
|
|
least 4 times. |
| 11 |
|
|
Options: |
| 12 |
|
|
-c sets the repeat count to <repeat#> (default 4) |
| 13 |
|
|
/; |
| 14 |
|
|
umask 0002; |
| 15 |
|
|
getopts('c:o:') || die($usage."\n"); |
| 16 |
|
|
my $outfile=$Getopt::Std::opt_o; |
| 17 |
|
|
my $c=$Getopt::Std::opt_c || 4; |
| 18 |
|
|
|
| 19 |
|
|
{ |
| 20 |
|
|
local $/="\n>"; |
| 21 |
|
|
while (<>) { |
| 22 |
|
|
s/^>//; |
| 23 |
|
|
chomp; |
| 24 |
|
|
my ($header, $seq)=(m/^([^\n]+)\n(.+)/s); |
| 25 |
|
|
$seq =~ tr/\t \n\r//d; |
| 26 |
|
|
|
| 27 |
|
|
$seq =~ s/((\w{2,}?)\2{$c,})/'N' x length $1/oeg; |
| 28 |
|
|
# for poly-nucleotide |
| 29 |
|
|
#$seq =~ s/((\w)\2{6,})/'N' x length $1/oeg; |
| 30 |
|
|
$seq =~ s/((\w)\2{5,})/'N' x length $1/eg; |
| 31 |
|
|
#print $seq."\n"; |
| 32 |
|
|
print ">$header\n"; |
| 33 |
|
|
print join("\n", unpack('(A70)*', $seq))."\n"; |
| 34 |
|
|
} |
| 35 |
|
|
} |
