Merge branch 'release3' of github.com:micheldumontier/bio2rdf-scripts into release3

Conflicts:
	irefindex/irefindex.php

Author: micheldumontier

README.md

@@ -36,7 +36,8 @@ class GOAParser extends Bio2RDFizer
 {
 	function __construct($argv) {
 		parent::__construct($argv,"goa");
-		parent::addParameter('files',true,'all|arabidopsis|chicken|cow|dicty|dog|fly|human|mouse|pdb|pig|rat|uniprot|worm|yeast|zebrafish','all','all or comma-separated list of files to process');
+//		parent::addParameter('files',true,'all|arabidopsis|chicken|cow|dicty|dog|fly|human|mouse|pdb|pig|rat|uniprot|worm|yeast|zebrafish','all','all or comma-separated list of files to process');
+		parent::addParameter('files',true,'all|arabidopsis|chicken|cow|dicty|dog|fly|human|mouse|pig|rat|worm|yeast|zebrafish','all','all or comma-separated list of files to process');
 		parent::addParameter('download_url',false,null,'ftp://ftp.ebi.ac.uk/pub/databases/GO/goa/');		
 		parent::initialize();
 	}
@@ -161,9 +162,13 @@ function process($file){
 
 			//entity id
 			$eid = $this->getdbURI($db,$id);
+if(!$eid) {
+print_r($fields);
+continue;
+}
 			parent::addRDF(
 				parent::describeIndividual($eid,$label,parent::getVoc()."GO-Annotation").
-				parent::describeClass(parent::getVoc()."GO-Annotation","GO Annotation").
+ 				parent::describeClass(parent::getVoc()."GO-Annotation","GO Annotation").
 				parent::triplifyString($eid,parent::getVoc()."symbol",$symbol)
 			);
 			parent::addRDF(
@@ -261,6 +266,8 @@ function getdbURI($db_id, $db_object_id){
 		} else if ($db_id == "PDB"){
 			$split_object = explode("_", $db_object_id);
 			$returnMe = "pdb:".$split_object[0]."/chain_".$split_object[1];
+		} else {
+			$returnMe = $db_id.":".$db_object_id;
 		}
 		return $returnMe;
 	}

goa/goa.php

@@ -117,11 +117,13 @@ function Run(){
 	function process(){
 		$header = $this->GetReadFile()->Read(200000);
 		$header_arr = explode("\t", $header);
-
-		if (count($header_arr) != 40)
+		$n = 41;
+		$c = count($header_arr);
+		if ($c != $n)
 		{
 			echo PHP_EOL;
-			trigger_error ("Header format is different than expected, please update the script",E_USER_ERROR);
+			print_r($header_arr);
+			trigger_error ("Expected $n columns, found $c . please update the script",E_USER_ERROR);
 			exit;
 		}
 
@@ -164,9 +166,10 @@ function process(){
 			$refseq_mappeddatasuppliedbyNCBI = $fields[34];
 			$uniprot_id_mappeddatasuppliedbyUniProt = $fields[35];
 			$ensembl_id_mappeddatasuppliedbyEnsembl = $fields[36];
-			$ucsc_id_mappeddatasuppliedbyUCSC = $fields[37];
-			$mouse_genome_database_id_mappeddatasuppliedbyMGI = $fields[38];
-			$rat_genome_database_id_mappeddatasuppliedbyRGD = $fields[39];
+			$vega_id_mappeddatasuppliedbyVega = $fields[37];
+			$ucsc_id_mappeddatasuppliedbyUCSC = $fields[38];
+			$mouse_genome_database_id_mappeddatasuppliedbyMGI = $fields[39];
+			$rat_genome_database_id_mappeddatasuppliedbyRGD = $fields[40];
 
 			$id_res = $id;
 			$id_label = "Gene Symbol for ".$approved_symbol;
@@ -464,6 +467,15 @@ function process(){
 				}
 			}
 
+			if(!empty($ucsc_id_mappeddatasuppliedbyVega)){
+				$ucsc_id_mappeddatasuppliedbyVega = explode(", ", $ucsc_id_mappeddatasuppliedbyVega);
+				foreach ($ucsc_id_mappeddatasuppliedbyVega as $vega_id) {
+					parent::AddRDF(
+						parent::triplify($id_res, $this->getVoc()."x-vega", "vega:".trim($vega_id)).
+						parent::describeProperty($this->getVoc()."x-vega", "Vega entry")
+					);
+				}
+			}
 			if(!empty($ucsc_id_mappeddatasuppliedbyUCSC)){
 				$ucsc_id_mappeddatasuppliedbyUCSC = explode(", ", $ucsc_id_mappeddatasuppliedbyUCSC);
 				foreach ($ucsc_id_mappeddatasuppliedbyUCSC as $ucsc_id) {