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

View of /FigKernelScripts/load_features.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.18 - (download) (as text) (annotate)
Thu Jun 9 05:51:06 2005 UTC (14 years, 6 months ago) by parrello
Branch: MAIN
CVS Tags: merge-trunktag-bobdev_news-2, Root-bobdev_news, merge-bobdev_news-1, merge-trunktag-bobdev_news-1, merge-bodev_news-3, merge-bobdev_news-2, merge-trunktag-bodev_news-3
Branch point for: Branch-bobdev_news
Changes since 1.17: +137 -201 lines
*** empty log message ***

# -*- perl -*-

use strict;
use FIG;
my $fig = new FIG;

use Tracer;

Trace("Preparing to load features.") if T(2);
my ($mode, @genomes) = FIG::parse_genome_args(@ARGV);
my $temp_dir = "$FIG_Config::temp";
my $organisms_dir = "$FIG_Config::organisms";

my($genome,@types,$type,$id,$loc,@aliases,$aliases,$contig);

# usage: load_features [G1 G2 G3 ... ]

Open(\*REL, ">$temp_dir/tmpfeat$$");
Open(\*ALIAS, "| sort -T $temp_dir -u > $temp_dir/tmpalias$$");


if ($mode eq 'all') {

    # Here we extract external aliases from the peg.synonyms table, when they can be inferred
    # accurately.
	Trace("Extracting external aliases from the peg.synonyms table.") if T(2);
    open(\*SYN, "<$FIG_Config::global/peg.synonyms");
    while (defined($_ = <SYN>))
    {
		chop;
		my($x,$y) = split(/\t/,$_);
		my @ids = map { $_ =~ /^([^,]+),(\d+)/; [$1,$2] } ($x,split(/;/,$y));
		my @fig = ();
		my(@nonfig) = ();
		foreach $_ (@ids)
		{
			if ($_->[0] =~ /^fig\|/)
			{
				push(@fig,$_);
			}
			else
			{
				push(@nonfig,$_);
			}
		}
	
		foreach $x (@fig)
		{
			my($peg,$peg_ln) = @$x;
			my $genome = &FIG::genome_of($peg);
			foreach $_ (@nonfig)
			{
				if ((@fig == 1) || ($peg_ln == $_->[1]))
				{
					print ALIAS "$peg\t$_->[0]\t$genome\n";
					Trace("Alias record $peg, $_->[0] for $genome.") if T(4);
				}
			}
		}
    }
    close(SYN);
}

my $changes = {};
if (open(TMP, "<$FIG_Config::global/changed.location.features"))
{
    while ($_ = <TMP>)
    {
		if ($_ =~ /^(fig\|\d+\.\d+\.[a-zA-Z]+\.\d+)/)
		{
			$changes->{$1}++;
		}
    }
    close(TMP);
}

foreach $genome (@genomes)
{
	Trace("Processing $genome.") if T(3);
    opendir(FEAT,"$organisms_dir/$genome/Features") 
		|| die "could not open $genome/Features";
    @types = grep { $_ =~ /^[a-zA-Z]+$/ } readdir(FEAT);
    closedir(FEAT);

    foreach $type (@types)
    {
		if ((-s "$organisms_dir/$genome/Features/$type/tbl") &&
			open(TBL,"<$organisms_dir/$genome/Features/$type/tbl"))
		{
			Trace("Loading $genome/Features/$type/tbl") if T(4);
			while (defined($_ = <TBL>))
			{
				chop;
				($id,$loc,@aliases) = split(/\t/,$_);
		
				if ($id && ($_ = $changes->{$id}))   # check for obsolete entries due to location changes
				{
					$changes->{$id}--;
					next;
				}
		
				if ($id)
				{
					my($minloc,$maxloc);
					if ($loc)
					{
						$loc =~ s/\s+$//;
						($contig,$minloc,$maxloc) = &FIG::boundaries_of($loc);
						if ($minloc && $maxloc)
						{
							($minloc < $maxloc) || (($minloc,$maxloc) = ($maxloc,$minloc));
						}
					}
		
					if (! $contig)
					{ 
						$loc = $contig = $minloc = $maxloc = ""; 
					}
		
					if (@aliases > 0)
					{
						$aliases = join(",",grep(/\S/,@aliases));
						my $alias;
						foreach $alias (@aliases)
						{
							if ($alias =~ /^([NXYZA]P_|gi\||sp\|\tr\||kegg\||uni\|)/)
							{
							
								print ALIAS "$id\t$alias\t$genome\tOVERRIDE\n";
								Trace("$id override alias $alias for $genome") if T(4);
							}
						}
					}
					else
					{
						$aliases = "";
					}
					$minloc = (! $minloc) ? 0 : $minloc;
					$maxloc = (! $maxloc) ? 0 : $maxloc;
					if ((length($loc) < 5000) && (length($contig) < 96) && (length($id) < 32) && ($id =~ /(\d+)$/))
					{
						print REL "$id\t$1\t$type\t$genome\t$loc\t$contig\t$minloc\t$maxloc\t$aliases\n";
					}
				}
			}
			close(TBL);
		}
    }
}
close(REL);
close(ALIAS);
Open(\*ALIASIN, "<$temp_dir/tmpalias$$");
Open(\*ALIASOUT, ">$temp_dir/tmpalias$$.1");
Trace("Parsing alias file.") if T(2);
$_ = <ALIASIN>;
while ($_ && ($_ =~ /^(\S+)/))
{
    my @aliases = ();
    my $curr = $1;
    while ($_ && ($_ =~ /^(\S+)\t(\S+)(\t(\S+))?/) && ($1 eq $curr))
    {
		push(@aliases,[$2,$3 ? 1 : 0]);
		$_ = <ALIASIN>;
    }
    my $x;
    my $genome = &FIG::genome_of($curr);
    foreach $x (@aliases)
    {
		if ($x->[1])
		{
			print ALIASOUT "$curr\t$x->[0]\t$genome\n";
		}
		else
		{
			my $i;
			for ($i=0; ($i < @aliases) && ((! $aliases[$i]->[1]) || (! &same_class($x->[0],$aliases[$i]->[0]))); $i++) {}
			if ($i == @aliases)
			{
				print ALIASOUT "$curr\t$x->[0]\t$genome\n";
			}
		}
    }
}
close(ALIASIN);
close(ALIASOUT);
unlink("$temp_dir/tmpalias$$");

$fig->reload_table($mode, 'features',
				   "id varchar(32), idN INTEGER, type varchar(16),genome varchar(16),"  .
			            "location TEXT,"  .
			            "contig varchar(96), minloc INTEGER, maxloc INTEGER,"  .
			            "aliases TEXT",
					{ features_id_ix => "id", features_org_ix => "genome",
					  features_type_ix => "type", features_beg_ix => "genome, contig, minloc" },
					"$temp_dir/tmpfeat$$", \@genomes);
unlink("$temp_dir/tmpfeat$$");

$fig->reload_table($mode, 'ext_alias',
					"id varchar(32), alias varchar(32), genome varchar(16)",
					{ ext_alias_alias_ix => "alias", ext_alias_genome_ix => "genome",
					  ext_alias_ix_id => "id" },
					"$temp_dir/tmpalias$$.1", \@genomes );

unlink("$temp_dir/tmpalias$$.1");
Trace("Features loaded.") if T(2);

sub same_class {
    my($x,$y) = @_;

    my $class1 = &classA($x);
    my $class2 = &classA($y);
    return ($class1 && ($class1 eq $class2));
}

sub classA {
    my($alias) = @_;

    if ($alias =~ /^([^\|]+)\|/)
    {
		return $1;
    }
    elsif ($alias =~ /^[NXYZA]P_[0-9\.]+$/)
    {
		return "refseq";
    }
    else
    {
		return "";
    }
}

MCS Webmaster
ViewVC Help
Powered by ViewVC 1.0.3