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

View of /FigKernelScripts/verify_pegs_in_tbl.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.1 - (download) (as text) (annotate)
Mon Dec 1 20:46:40 2003 UTC (16 years, 4 months ago) by efrank
Branch: MAIN
CVS Tags: V00-00-01, merge-trunktag-bobdev_news-2, Root-bobdev_news, merge-bobdev_news-1, caBIG-dataload-0, merge-trunktag-bobdev_news-1, merge-bodev_news-3, efrank-preMakefileToStandardMk, delong-01, delong-02, caBIG-00-00-00, merge-bobdev_news-2, V00-00-00, merge-trunktag-bodev_news-3
Branch point for: Branch-bobdev_news

Makefile:
	get it to work with the release tools

All the rest:
	had to rename foo to foo.pl so that makefiles could recognize
	perl source from, say, Makefiles and READMEs

use FIG;


# usage: verify_pegs_in_tbl Contigs < tbl > report

($contigs = shift @ARGV)
    || die "usage: verify_pegs_in_tbl Contigs < tbl > report";

open(CONTIGS,$contigs) || die "could not open $contigs";
$/ = "\n>";
while (defined($_ = <CONTIGS>))
{
    chomp;
    if ($_ =~ /^>?(\S+)[^\n]*\n(.*)/s)
    {
	$id  =  $1;
	$seq =  $2;
	$seq =~ s/\s//g;
	$contigs{$id} = uc $seq;
    }
}
close(CONTIGS);
$/ = "\n";

while (defined($_ = <STDIN>))
{
    if (($_ =~ /^(fig\|\d+\.\d+\.peg\.\d+)\t(\S+)/) &&
	(($rid,$loc) = ($1,$2)) &&
	($loc =~ /^([^,]+)_(\d+)_\d+(,|$)/) &&
	(($contig,$beg) = ($1,$2)) &&
	($loc =~ /$contig\_\d+_(\d+)$/) &&
	($end = $1))
    {
	$seq = &FIG::extract_seq(\%contigs,$loc);
	if ($contigs{$contig} && ($beg < $end))
	{
	    $start = substr($contigs{$contig},$beg-1,3);
	    $stop  = substr($contigs{$contig},$end,3);
	    for ($i=0; ($i < (length($seq) - 3)) && (substr($seq,$i,3) !~ /(TAG|TGA|TAA)/); $i += 3) {}
	    if ($i < (length($seq)-3))
	    {
		print "$rid has an embedded stop codon\n";
	    }
	}
	elsif ($contigs{$contig})
	{
	    $start = &reverse_comp(substr($contigs{$contig},$beg-3,3));
	    $stop  = &reverse_comp(substr($contigs{$contig},$end-4,3));
	    for ($i=0; ($i < (length($seq) - 3)) && (substr($seq,$i,3) !~ /(TAG|TGA|TAA)/); $i += 3) {}
	    if ($i < (length($seq)-3))
	    {
		print "$rid has an embedded stop codon\n";
	    }
	}
	else
	{
	    print STDERR "invalid: $_";
	}

	if (($start !~ /^[ATG]TG/) && ($beg > 40) && ($beg < (length($contigs{$contig})-40)))
	{
	    print "bad start for $rid: $start\n";
	}
	if (($stop !~ /(TAG|TGA|TAA)/) && ($end > 40) && ($end < (length($contigs{$contig}) - 40)))
	{
	    print "bad stop for $rid: $stop\n";
	}
	$strand = ($beg < $end) ? "+" : "-";

	push(@{ $on{$contig} }, [&FIG::min($beg,$end),&FIG::max($beg,$end),$strand,$rid]);
    }
}

foreach $contig (sort keys(%on))
{
    $x = $on{$contig};
    @on = sort { ($a->[0] <=> $b->[0]) or ($a->[1] <=> $b->[1]) or ($a->[2] cmp $b->[2]) } @$x;

    for ($i=0; ($i < $#on); $i++)
    {
	for ($j=$i+1; ($j < @on) && ($on[$i]->[1] > ($on[$j]->[0]) + 10); $j++)
	{
	    $overlap = ($on[$j]->[1] > $on[$i]->[1]) ?
		           ($on[$i]->[1] - $on[$j]->[0]) + 1 :
			   ($on[$j]->[1] - $on[$j]->[0]) + 1;
	    push(@overlaps,[$overlap,"$overlap\t$on[$i]->[2],$on[$j]->[2]\t$on[$i]->[3] overlaps $on[$j]->[3]\n"]);
	}
    }
}

foreach $_ (sort { $b->[0] <=> $a->[0] } @overlaps)
{
    print $_->[1];
}
	
sub reverse_comp {
    my( $seq ) = @_;
    my( $rev  );

    $rev =  reverse( $seq );
    $rev =~ tr/a-z/A-Z/;
    $rev =~ tr/ACGTUMRWSYKBDHV/TGCAAKYWSRMVHDB/;
    return $rev;
}

MCS Webmaster
ViewVC Help
Powered by ViewVC 1.0.3