view VCF_to_VariantDB.pl @ 9:3b27cae9b359 draft

New Version. VariantDB checks for Genome Build matching using the Galaxy dbkey variable.

#!/usr/bin/perl

# load modules
use Getopt::Std;

##########################
# COMMAND LINE ARGUMENTS #
##########################
# v = (v)cf file to load
# V = (V)CF file encoded id
# u = (u)ser email from galaxy
# n = sample (n)ame
# a = sample (a)nnotation
# g = sample (g)ender
# o = (o)utput file (simple text file)
# b = (b)am file (optional)
# B = (B)am index , needed if b is specified
# c = encoded id of bam file (optional)
# C = encoded id of Bam index , needed if b is specified => NOT POSSIBLE YET, NEEDS INDEXING ON VARIANTDB SERVER !
# S = (S)erver addrress to send data to.
# R = (r)oot of galaxy web server (/home/galaxyuser/galaxy-dist)
# H = (H)ost of the galaxy web server (http://my.galaxy.server/galaxy/)
# G = Genome build (dbkey)
# F = (F)ormat of input file 
getopts('v:u:n:a:g:o:b:B:V:c:S:R:H:G:F:', \%opts);  # option are in %opts
$|++;
#################
## CHECK INPUT ##
#################
if (!exists($opts{'v'})) {
	die('No VCF File Specified');
}
if (!-e $opts{'v'}) {
	die('VCF File not found');
}
if (!exists($opts{'u'})) {
	die('No user specified');
}
if (!exists($opts{'S'})) {
	die('No VariantDB server specified');
}
if (!exists($opts{'H'})) {
	die('The Galaxy source-server is not specified');
}
if (!exists($opts{'F'})) {
	die('VCF format not specified.');
}

################
# open outfile #
################
open OUT, ">$opts{'o'}";

###############################
## TEST CONNECTION TO SERVER ##
###############################
use LWP::UserAgent;
my $url = $opts{'S'}."/";
$url =~ s/\/\/$/\//;
$url .= "cgi-bin/galaxy_communication.cgi";
my $conn = LWP::UserAgent->new();
$conn->timeout(1800);
my $response = $conn->post( $url, {'HelloWorld' => 1} );
my $content = $response->decoded_content();

if ($content eq 'HelloGalaxy') {
	print OUT "Testing connection to $opts{'S'} : OK.\n";
}
else {
	die("Could not connect to the specified server : $content");
}
	

##################
## TEST USER ID ##
##################
$email = $opts{'u'};
$dbkey = $opts{'G'};
my $response = $conn->post( $url, {'CheckUser' => $email,'dbkey' => $dbkey} );
my $content = $response->decoded_content();

if ($content eq 'OK') {
	print OUT "Testing User-existence and Genome Build: OK.\n";
}
else {
	die("ERROR: $content");
}

print $opts{'H'}. " was specified as galaxy host\n";

###############################################
## SEND THE VCF AND BAM FILES FOR PROCESSING ##
###############################################
# filepaths
my $vcfpath = $opts{'v'};
my $bampath = $opts{'b'};
my $baipath = $opts{'B'};
# input VCF format
$format = $opts{'F'};
# make output directory in (galaxy/static/) working dir
my $rand = int(rand(1000));
our $wd = $opts{'R'}."/static/VCF_parser.".$rand; #int(rand(1000));
our $dd = $opts{'H'}."/static/VCF_parser.".$rand;
while (-d $wd) {
	my $rand = int(rand(1000));
	$wd = $opts{'R'}."/static/VCF_parser.".$rand;#int(rand(1000));
	$dd = $opts{'H'}."/static/VCF_parser.".$rand;

}
$result = system("mkdir $wd");


## link files
$vcfurl = "$dd/data.vcf";
$vcfurl =~ s/\s//g;
system ("ln -s $vcfpath $wd/data.vcf");
if (exists($opts{'b'})) {
	$bamurl = "$dd/data.bam";
	$bamurl =~ s/\s//g;
	$bamidxurl = "$dd/data.bai";
	$bamidxurl =~ s/\s//g;
	system ("ln -s $bampath $wd/data.bam");
	system ("ln -s $baipath $wd/data.bai");
}	
$sample = $opts{'n'};
$gender = $opts{'g'};
# post form to the variantDB host. 
if (exists($opts{'b'})) {
	$response = $conn->post( $url, {'VCFurl1' => "$vcfurl", 'BAMurl1' => "$bamurl", 'BAIurl1' => "$bamidxurl",'storedata1' => 1, 'name1' => "$sample", 'gender1' => "$gender", 'User' => $email, 'GalaxyUpload' => 1,'Format1' => $format} );
}
else {
	$response = $conn->post( $url, {'VCFurl1' => "$vcfurl",  'name1' => "$sample", 'gender1' => "$gender", 'User' => $email, 'GalaxyUpload' =>1, 'Format1' => $format } );
}
my $content = $response->decoded_content();
chomp($content);
## check if upload went ok.
if (substr($content,0,2) ne 'OK') {
	die("ERROR: $content");
}
## extract wd from content.
print OUT "Uploading datafiles to VariantDB : OK.\n";
$content =~ m/OK-(.+)$/;
$rwd = $1;
if ($rwd eq '') {
	die("ERROR : No remote working directory provided to check status.");
}
## now wait for the import to finish.
$status = 0;
while ($status == 0) {
	my $response = $conn->post( $url, {'CheckStatus' => 1,'rwd' => $rwd}) ;
	$content = $response->decoded_content();
	chomp($content);
	if (substr($content,0,2) ne 'OK') {
		die("ERROR: $content");
	}
	$status = substr($content,3,1);
	sleep 10;
}

## Loading OK
# latest respons : OK-1-Content
$content = substr($content,5);
print OUT "Processing Datafiles : OK.\n";
print OUT "\n$content\n";
close OUT;

# clean up 
system("rm -Rf '$wd'");