[Bio] / FigKernelScripts / reformat_fasta.pl Repository:
ViewVC logotype

View of /FigKernelScripts/reformat_fasta.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.1 - (download) (as text) (annotate)
Wed Sep 28 21:41:01 2005 UTC (14 years, 8 months ago) by overbeek
Branch: MAIN
CVS Tags: caBIG-dataload-0, caBIG-00-00-00
Script to standardize FASTA format files. -- /gdp

# -*- perl -*-

use FIG;

use Pod::Text;
# use File::Basename;   $this_tool_name = basename($0);
if ((@ARGV == 1) && ($ARGV[0] =~ m/-help/))  {
    pod2text($0);  exit(0);
}

=pod

=over 5

=item Usage:     reformat_fasta_file  < fasta_in  > fasta_out

=item Function:  Reformats a fasta file to a uniform 60 chars per line.

=back

=cut

if ($ARGV[0] =~ m/-strip/) { $strip = shift; }

if ($ARGV[0] && (-e $ARGV[0]))
{
    open(FIN, "<$ARGV[0]") or die "could not open $ARGV[0] as input";
    $fin = \*FIN;
}
else
{
    $fin = \*STDIN;
}

if ($ARGV[1])
{
    open(FOUT, ">$ARGV[1]") or die "could not open $ARGV[1] as output";
    $fout = \*FOUT;
}
else
{
    $fout = \*STDOUT;
}



while ( (! eof($fin)) && ( ($head, $seq) = &get_a_fasta_record($fin) ) )
{
    if ($strip) { $head =~ m/^(\S+)/;  $head = $1; }
    
    &display_id_and_seq( $head, \$seq, $fout );
}


sub get_a_fasta_record
{
    my ($fh) = @_;
    my ( $old_eol, $entry, @lines, $head, $id, $seq ); 

    $old_eol = $/;
    $/ = "\n>";
    
    if (! defined($fh))  { $fh = \*STDIN; }
    
    $entry = <$fh>;
    chomp $entry;
    @lines =  split( /\n/, $entry );
    
    $head  =  shift @lines;
    $head  =~ s/^>?//;
    $head  =~ m/(\S+)/;   $id = $1;
    
    $seq   =  join( "", @lines );
    $seq   =~ s/\s//g;
    $seq   =~ s/\*$//;
    
    my $tmp = $seq;
    if (($tmp =~ s/[ARNDCQEGHILKMFPSTWYUVBZX]//ig) && $tmp)
    {
	print STDERR "$id has invalid chars: $tmp\n";
	$seq  =~ s/[^ARNDCQEGHILKMFPSTWYUVBZX]/x/ig;
    }
    
    $/ = $old_eol;
    return ( $head, $seq );
}


sub display_id_and_seq {
    my( $id, $seq, $fh ) = @_;
    my ( $i, $n, $ln );
    
    if (! defined($fh) )  { $fh = \*STDOUT; }
    
    print $fh ">$id\n";
    
    $n = length($$seq);
#   confess "zero-length sequence ???" if ( (! defined($n)) || ($n == 0) );
    for ($i=0; ($i < $n); $i += 60)
    {
	if (($i + 60) <= $n)
	{
	    $ln = substr($$seq,$i,60);
	}
	else
	{
	    $ln = substr($$seq,$i,($n-$i));
	}
	print $fh "$ln\n";
    }
}

MCS Webmaster
ViewVC Help
Powered by ViewVC 1.0.3