Merge pull request #444 from micheldumontier/master

changes around the board

Author: micheldumontier

bioportal/bioportal.php

@@ -394,7 +394,7 @@ function OBO2RDF($abbv)
 		$graph_uri = '<'.parent::getRegistry()->getFQURI(parent::getGraphURI()).'>';
 		$bid = 1;
 
-		while($l = parent::getReadFile()->read()) {
+		while(FALSE !== ($l = parent::getReadFile()->read())) {
 			$lt = trim($l);
 			if(strlen($lt) == 0) continue;
 			if($lt[0] == '!') continue;
@@ -676,7 +676,8 @@ function OBO2RDF($abbv)
 			} else {
 				//header
 				//format-version: 1.0
-				$buf .= parent::triplifyString($ouri,"obo_vocabulary:$a[0]",str_replace( array('"','\:'), array('\"',':'), isset($a[1])?$a[1]:""));
+				$buf .= parent::triplifyString($ouri,"obo_vocabulary:$a[0]",
+					str_replace( array('"','\:'), array('\"',':'), isset($a[1])?$a[1]:""));
 			}
 
 			if($minimal || $minimalp) parent::getWriteFile()->write($min);

ctd/ctd.php

@@ -420,13 +420,13 @@ function CTD_chem_pathways_enriched()
 				return FALSE;
 			}
 			$first = false;
-		}
-		
+		}	
 		$chemical_id = $a[1];
 
 		$this->getRegistry()->parseQName($a[4], $pathway_ns, $pathway_id);
 		if($pathway_ns == "react") $pathway_ns = "reactome";
-
+		if($pathway_ns == "kegg") $pathway_id = "map".$pathway_id;
+		
 		$pathway_resource_id = parent::getRes().md5($chemical_id.$pathway_ns.$pathway_id.$a[6]);
 		$pathway_resource_label = "Chemical-pathway association between mesh:".$chemical_id." and ".$pathway_ns.":".$pathway_id." with p-value ".$a[6];
 
@@ -509,7 +509,8 @@ function CTD_diseases_pathways()
 		$this->getRegistry()->parseQName($a[1],$disease_ns,$disease_id);
 		$this->getRegistry()->parseQName($a[3],$pathway_ns,$pathway_id);
 		if($pathway_ns == 'react') $pathway_ns = 'reactome';
-
+		if($pathway_ns == "kegg") $pathway_id = "map".$pathway_id;
+		
 		$this->AddRDF(
 			parent::triplify($disease_ns.":".$disease_id, $this->getVoc()."pathway", $pathway_ns.":".$pathway_id).
 			parent::triplifyString($disease_ns.":".$disease_id, "rdfs:label", $a[0]." [$disease_ns:$disease_id]").
@@ -612,6 +613,7 @@ function CTD_genes_pathways()
 		$this->getRegistry()->parseQName($a[3],$pathway_ns,$pathway_id);
 		$pathway_id = trim($pathway_id);
 		if($pathway_ns == "react") $pathway_ns = "reactome";
+		if($pathway_ns == "kegg") $pathway_id = "map".$pathway_id;
 
 		$this->ADDRDF(
 			parent::triplify($gene_ns.":".$gene_id, $this->getVoc()."pathway", $pathway_ns.":".$pathway_id).
@@ -645,7 +647,8 @@ function CTD_Pathways()
 
 		$this->getRegistry()->parseQName(trim($a[1]),$pathway_ns,$pathway_id);	
 		if($pathway_ns == "react") $pathway_ns = "reactome";	
-
+		if($pathway_ns == "kegg") $pathway_id = "map".$pathway_id;
+		
 		$this->AddRDF(
 			parent::describeIndividual($pathway_ns.":".$pathway_id, $a[0], $this->getVoc()."Pathway").
 			parent::describeClass($this->getVoc()."Pathway", "CTD Pathway")	

drugbank/drugbank.php

@@ -56,6 +56,16 @@ function Run()
 		if(parent::getParameterValue("id_list")) {
 			$this->id_list = array_flip(explode(",",parent::getParameterValue('id_list')));
 		}
+		
+		$go_cache_file = parent::getParameterValue('indir')."go.cache.json";
+		#unlink($go_cache_file);
+		if(!file_exists($go_cache_file) or parent::getParameterValue('download') == true) {
+			$this->getGO();
+			file_put_contents($go_cache_file,json_encode($this->go));
+		} else {
+			// read the file
+			$this->go = json_decode( file_get_contents($go_cache_file), true);		
+		}
 
 		$dataset_description = '';
 		foreach($files AS $f) {
@@ -241,9 +251,16 @@ function parsePartnerEntry($did, $pid, $x)
 						parent::triplify($pid, parent::getVoc()."x-pfam","pfam:"."".$v2->identifier)
 					);
 				} else if($k2 == "go-classifier") {
-					parent::addRDF(
-						parent::triplifyString($pid, parent::getVoc()."go-".$v2->category, $v2->description)
-					);
+					$e = array_search($v2->description, $this->go);
+					if($e !== FALSE) {
+						parent::addRDF(		
+							parent::triplify($pid, parent::getVoc()."go-".$v2->category, $e)
+						);				
+					} else {
+						parent::addRDF(				
+							parent::triplifyString($pid, parent::getVoc()."go-".$v2->category, $v2->description)
+						);
+					}
 				} else {
 					trigger_error("no handler for $k2",E_USER_WARNING);
 	/*					parent::addRDF(
@@ -773,6 +790,35 @@ function AddList(&$x, $id, $list_name, $item_name, $predicate, $list_item_name =
 		}
 	}
 
+	function getGO() 
+	{
+		$this->go = null;
+		
+		$server = "http://bio2rdf.org/sparql";
+		$sparql = "PREFIX dct: <http://purl.org/dc/terms/>
+SELECT distinct ?id ?title
+{
+ ?go a <http://bio2rdf.org/go_vocabulary:Resource> .
+ ?go dct:identifier ?id.
+ ?go dct:title ?title .
+} ";
+		$url = $server."?query=".urlencode($sparql)."&format=".urlencode("text/tab-separated-values");
+
+		$results = file_get_contents($url);
+		if($results === FALSE) {
+			trigger_error("Unable to get Gene Ontology labels",E_USER_WARNING);
+			return false;
+		}
+		$list = explode("\n",$results);
+		array_shift($list); array_pop($list);  // remove first and last
+		
+		foreach($list AS $v) {
+			$b = explode("\t",str_replace('"','',$v));
+			$this->go[$b[0]] = $b[1];
+		}
+		return true;
+	}
+	
 } // end class
 
 ?>

kegg/kegg.php

@@ -59,7 +59,8 @@ function run()
 
 		// handle genes separately
 		if(in_array("genes",$files)) {	
-			$orgs = array("hsa"); //,"mmu","eco","dre","dme","ath","sce","ddi");
+			$orgs = array("hsa","mmu","eco","dre","dme","ath","sce","ddi");
+			//$orgs = array("hsa"); 
 			echo "processing genes".PHP_EOL;
 
 			$ofile = "kegg-genes.".parent::getParameterValue('output_format'); 
@@ -82,7 +83,7 @@ function run()
 
 				// get the list of genes for this organims
 				echo "processing $org".PHP_EOL;
-				$this->org = $org; // local variable
+				$this->org = strtoupper($org); // local variable
 
 				$lfile = $ldir.$org.".txt";
 				$rfile = parent::getParameterValue("download_url")."list/$org";
@@ -223,9 +224,10 @@ function process($db)
 			if(isset($this->idlist) and !in_array($id,$this->idlist)) continue;
 
 			if(isset($this->org)) {
-				$id = $ns."_".$id;
+				$id = strtoupper($ns)."_".$id;
 			}
 			$uri = $this->getNamespace().$id;
+
 			parent::addRDF(
 				parent::describeIndividual($uri,$name,parent::getVoc().ucfirst($db)).
 				parent::describeClass(parent::getVoc().ucfirst($db),"KEGG $db").
@@ -293,6 +295,7 @@ function parseEntry($lfile)
 				$uri = parent::getNamespace().$e['id'];
 				continue;
 			}
+			
 			// key with value
 			if(in_array($k, array("NAME","DESCRIPTION","DEFINITION","EQUATION","COMMENT"))) {
 				if($k == "NAME") {
@@ -688,14 +691,19 @@ function parseEntry($lfile)
 					parent::triplify($uri,parent::getVoc().strtolower($k),$id)
 				);
 				preg_match_all("/ \[([^\]]+)\]/",$v,$m);
-				if(isset($m[1])) {
-					foreach($m[1] AS $item) {
-						if(!strstr($item,"KO")) $item = "kegg:".str_replace(":","_",$item);
-						else $item = str_replace("KO:","kegg:",$item);
-						parent::addRDF(
-							parent::triplify($id,parent::getVoc()."link",$item)
-						);
+				if(isset($m[1]) and !empty($m[1])) {					
+					foreach($m[1] AS $item) {			
+						$a = explode(':',$item);  // get the namespace
+						$b = explode(' ',$a[1]);
+						foreach($b AS $c) {
+							if(!strstr($item,"KO")) $i = "kegg:".$a[0].'_'.$c;
+							else $i = "kegg:".$c;
+							parent::addRDF(
+								parent::triplify($id,parent::getVoc()."link",$i)
+							);
+						}
 					}
+					$test = true;
 				}
 				continue;
 			}