Merge pull request #475 from micheldumontier/master

updates for v5

Author: micheldumontier

Diff for: bioportal/bioportal.php

@@ -367,7 +367,7 @@ public function TriplifyMap($a, $prefix)
 
 		} else {
 			parent::addRDF(
-				parent::triplifyString($s_uri,$p_uri,addslashes($a['o']),(($a['o_datatype'] == '')?null:$a['o_datatype']),(($a['o_lang'] == '')?null:$a['o_lang']))
+				parent::triplifyString($s_uri,$p_uri,$a['o'],(($a['o_datatype'] == '')?null:$a['o_datatype']),(($a['o_lang'] == '')?null:$a['o_lang']))
 			);			
 		}
 
@@ -464,7 +464,7 @@ function OBO2RDF($abbv)
 					$tid = $ns.":".$id;
 					echo $tid.PHP_EOL;
 				} else if($a[0] == "name") {
-					$name = addslashes(stripslashes($a[1]));
+					$name = stripslashes($a[1]);
 					$buf .= parent::describeClass($tid,$name);
 					$buf .= parent::triplifyString($tid,"dc:title",$name);
 				} else if($a[0] == "is_a") {
@@ -520,7 +520,9 @@ function OBO2RDF($abbv)
 						$b = explode(":",$a[1],2);
 						if(isset($b[1])) {
 							if(substr($b[1],0,4) == "http") {
-								$buf .= parent::triplify($tid,"rdfs:seeAlso", stripslashes($b[1]));
+								// https://en.wikipedia.org/wiki/Prolamin {source="SUBMITTER"}
+								$url = preg_replace("/{.*\}/","",$b[1]);
+								$buf .= parent::triplify($tid,"rdfs:seeAlso", $url);
 							} else {
 								$ns = str_replace(array(" ","\\",) ,"",strtolower($b[0]));
 								$id = trim($b[1]);
@@ -547,10 +549,12 @@ function OBO2RDF($abbv)
 								if($ns == "submitter") $ns = "chebi.submitter";
 								if($ns == "wikipedia" || $ns == "mesh") $id = str_replace(" ","+",$id);
 								if($ns == "id-validation-regexp") {
-									$buf .= parent::triplifyString($tid,"obo_vocabulary:$ns", addslashes($id));
+									$buf .= parent::triplifyString($tid,"obo_vocabulary:$ns", $id);
 								} else {
-									if($ns) 
-										$buf .= parent::triplify($tid,"obo_vocabulary:x-$ns", "$ns:".str_replace(" ","-",$id));
+									if($ns) {
+										$id = str_replace(array(" ",",","#","<",">"),array("%20","%2C","%23","%3C","%3E"),$id);
+										$buf .= parent::triplify($tid,"obo_vocabulary:x-$ns", "$ns:$id");
+									}
 								}
 							}
 						}
@@ -688,7 +692,7 @@ function OBO2RDF($abbv)
 				//header
 				//format-version: 1.0
 				$buf .= parent::triplifyString($ouri,"obo_vocabulary:$a[0]",
-					str_replace( array('"','\:'), array('\"',':'), isset($a[1])?$a[1]:""));
+					str_replace( array('\:'), array(':'), isset($a[1])?$a[1]:""));
 			}
 
 			if($minimal || $minimalp) parent::getWriteFile()->write($min);

Diff for: drugbank/drugbank.php

@@ -185,7 +185,11 @@ function NSMap($source)
             case 'genbank protein database':
                 return 'gi';
             case 'hugo gene nomenclature committee (hgnc)':
-                return 'hgnc';
+				return 'hgnc';
+			case 'therapeutic targets database':
+				return 'ttd';
+			case 'rxcui':
+				return 'rxnorm';
 
             default:
                 return strtolower($source);
@@ -455,21 +459,24 @@ function parseDrugEntry(&$xml)
 			foreach($x->mixtures->mixture AS $item) {
 				if(isset($item)) {
 					$o = $item;
-					$mid = parent::getRes().md5(str_replace(" ","-",$o->name[0]));
+					$label = str_replace(array(" ","%"),array("-", "percent"), $o->name[0]);
+					$ingredients = str_replace(array(" ","%"),array("-", "percent"), $o->ingredients[0]);
+					$mid = parent::getRes().md5($label);
 
 					parent::addRDF(
 						parent::triplify($did,parent::getVoc()."mixture",$mid).
-						parent::describeIndividual($mid,$o->name[0],parent::getVoc()."Mixture").
+						parent::describeIndividual($mid,$label,parent::getVoc()."Mixture").
 						parent::describeClass(parent::getVoc()."Mixture","mixture").
-						parent::triplifyString($mid,$this->getVoc()."ingredients","".$o->ingredients[0]) 
+						parent::triplifyString($mid,$this->getVoc()."ingredients","".$ingredients) 
 					);
 
 					$a = explode(" + ",$o->ingredients[0]);
 					foreach($a AS $b) {
 						$b = trim($b);
-						$iid = parent::getRes().str_replace(" ","-",$b);
+						$label = str_replace(array(" ",",","%"),array("-","-","percent"),$b);
+						$iid = parent::getRes().$label;
 						parent::addRDF(
-							parent::describeClass($iid,$b, parent::getVoc()."Ingredient").
+							parent::describeClass($iid,$label, parent::getVoc()."Ingredient").
 							parent::describeClass(parent::getVoc()."Ingredient","Ingredient").
 							parent::triplify($mid,parent::getVoc()."ingredient",$iid)
 						);
@@ -661,7 +668,7 @@ function parseDrugEntry(&$xml)
 					$ns = $this->NSMap($obj->resource);
 					$id = $obj->identifier;
 					if($ns == "genecards") $id = str_replace(array(" "),array("_"),$id);
-
+					if($ns == "wikipedia") $id = str_replace(array(","), "-", $id);
 					parent::addRDF(
 						parent::triplify($did,parent::getVoc()."x-$ns","$ns:$id")
 					);
@@ -781,7 +788,10 @@ function AddList(&$x, $id, $list_name, $item_name, $predicate, $list_item_name =
 		if(isset($mylist)) {
 			foreach($mylist AS $item) {
 				$label = ''.$item;
-				$kid = parent::getVoc().ucfirst(str_replace(" ","-",$label)); // generate a new identifier for the list item
+				$label = str_replace(array(" ","[","]",","),"-",$label);
+				$label = rawurlencode(utf8_encode($label));
+
+				$kid = parent::getVoc().ucfirst($label); // generate a new identifier for the list item
 				$this->addRDF(
 					$this->describeIndividual($kid,$label,parent::getVoc().ucfirst($item_name)).
 					$this->describeClass(parent::getVoc().ucfirst($item_name),ucfirst($item_name)).

Diff for: lsr/lsr.php

@@ -173,7 +173,7 @@ function parse()
 			if($r['pubmed']) {
 				foreach(explode(",",$r['pubmed']) AS $pubmed) {
 					parent::addRDF(
-						parent::triplify($id,"cito:citesAsAuthority","pubmed:".$pubmed)
+						parent::triplify($id,"cito:citesAsAuthority","pubmed:".trim($pubmed))
 //						parent::triplify("pubmed:".$pubmed, "rdf:type", "pubmed_vocabulary:Resource")
 					);
 				}
@@ -206,7 +206,7 @@ function parse()
 				&& $r['homepage'] !== 'dead'
 				&& $r['homepage'] !== 'unavailable') {
 				parent::addRDF(
-					parent::triplify($id,"foaf:page",$r['homepage'])
+					parent::triplify($id,"foaf:page",trim($r['homepage']))
 				);
 			}
 			if($r['license']) {

Diff for: mgi/mgi.php

@@ -266,14 +266,14 @@ function MGI_GenePheno()
 			}
 
 			if($a[5]) {
-				$pmids = explode(",",$a[5]);
+				$pmids = explode("|",$a[5]);
 				foreach($pmids AS $pmid) {
 					parent::addRDF(		
 						parent::triplify($id,$this->getVoc()."x-pubmed","pubmed:".$pmid)	
 					);
 				}
 			}
-			$b = explode(",",$a[6]);
+			$b = explode("|",$a[6]);
 			foreach($b AS $marker) {
 				parent::addRDF(
 					parent::triplify($id,$this->getVoc()."marker",$marker).

Diff for: pharmgkb/pharmgkb.php

@@ -370,11 +370,17 @@ function genes()
 		}
 	}
 
-	function parseList($str)
+	function parseList($str, $delim = ';')
 	{
 		$list = '';
-		if($str[0] == '"') $list = explode('","', substr($str,1,-1));
-		else $list = array($str);
+		if($str[0] == '"') {
+			$list = explode('","', substr($str,1,-1));
+		} else {
+			if(strstr($str,$delim)) {
+				$list = explode($delim, $str);
+			}
+		}
+		if(!is_array($list)) $list = array($str);
 		return $list;
 	}
 
@@ -454,7 +460,7 @@ function drugs()
 			if(trim($a[2])) { 
 				// generic names
 				// Entacapona [INN-Spanish],Entacapone [Usan:Inn],Entacaponum [INN-Latin],entacapone
-				$list = $this->parseList(trim($a[2]));
+				$list = $this->parseList(trim($a[2]),",");
 				foreach($list AS $c) {
 					parent::addRDF(
 						parent::triplifyString($id, parent::getVoc()."generic_name",  str_replace('"','',$c))
@@ -467,7 +473,7 @@ function drugs()
 			if(trim($a[3])) { 
 				// trade names
 				//Disorat,OptiPranolol,Trimepranol
-				$list = $this->parseList(trim($a[3]));
+				$list = $this->parseList(trim($a[3]),",");
 				foreach($list as $c) {
 					parent::addRDF(
 						parent::triplifyString($id, parent::getVoc()."trade_name", str_replace(array("'", "\""), array("\\\'", "") ,$c))
@@ -480,7 +486,7 @@ function drugs()
 			if(trim($a[4])) {
 				// Brand Mixtures	
 				// Benzyl benzoate 99+ %,"Dermadex Crm (Benzoic Acid + Benzyl Benzoate + Lindane + Salicylic Acid + Zinc Oxide + Zinc Undecylenate)",
-				$list = $this->parseList(trim($a[4]));
+				$list = $this->parseList(trim($a[4]),",");
 				foreach($list as $c) {
 					parent::addRDF(
 						parent::triplifyString($id, parent::getVoc()."brand_mixture", str_replace(array("'", "\""),array("\\\'",""), $c))
@@ -500,7 +506,7 @@ function drugs()
 			if(trim($a[6])) {
 				// Cross References	
 				// drugBank:DB00789,keggDrug:D01707,pubChemCompound:55466,pubChemSubstance:192903,url:http://en.wikipedia.org/wiki/Gadopentetate_dimeglumine
-				$list = $this->parseList(trim($a[6]));
+				$list = $this->parseList(trim($a[6]),",");
 				foreach($list as $c) {
 					$this->getRegistry()->parseQName($c,$ns,$id1);
 					if($ns == "chebi") $id1 = substr($id1, 6);
@@ -544,7 +550,10 @@ function drugs()
 				// External Vocabulary
 				// ATC:H01AC(Somatropin and somatropin agonists),ATC:V04CD(Tests for pituitary function)
 				// ATC:D07AB(Corticosteroids, moderately potent (group II)) => this is why you don't use brackets and commas as separators.
-				$list = $this->parseList(trim($a[10]));
+				$list = $this->parseList(trim($a[10]),",");
+				if(strstr($a[10],"potent")) { $c = array(implode(",",$list));$list = $c;}
+				else if(strstr($a[10],"weak")) { $c = array(implode(",",$list));$list = $c;}
+
 				foreach($list as $c) {
 					preg_match("/([^\(]+)?\((.*)\)/", $c, $m);
 					if(isset($m[1])) {				
@@ -566,7 +575,7 @@ function drugs()
 			}
 			if(trim($a[22])) {
 				// ATC identifiers
-				$list = $this->parseList(trim($a[22]));
+				$list = $this->parseList(trim($a[22]),",");
 				foreach($list as $c) {
 					parent::addRDF(
 						parent::triplify($id, parent::getVoc()."x-atc", "atc:".$c)
@@ -802,7 +811,7 @@ function variants()
 
 	function clinical_ann_metadata()
 	{
-		$header = array("Clinical Annotation Id","Location","Gene","Level of Evidence","Clinical Annotation Types","Genotype-Phenotype IDs","Annotation Text","Variant Annotations IDs","Variant Annotations","PMIDs","Evidence Count","Related Drugs","Related Diseases","Biogeographical groups", "Chromosome");
+		$header = array("Clinical Annotation Id","Location","Gene","Level of Evidence","Clinical Annotation Types","Genotype-Phenotype IDs","Annotation Text","Variant Annotations IDs","Variant Annotations","PMIDs","Evidence Count","Related Chemicals","Related Diseases","Biogeographical groups", "Chromosome","Latest History");
 		$this_header = explode("\t",$this->getReadFile()->read());
 		if(count($this_header) != count($header)) {
 			trigger_error("Change in the number of columns. Expected ".count($header).", but found ".count($this_header),E_USER_ERROR);
@@ -812,19 +821,8 @@ function clinical_ann_metadata()
 			$a = explode("\t",$l);
 
 			$id = parent::getNamespace().$a[0];
-			# fixing bad file formatting
-			if($a[0] == "982040598" or $a[0] == "982037603") {
-				$a[8] .= $a[11];
-				$a[9] = $a[12];
-				$a[10] = $a[13]; 
-				$a[11]= $a[14]; 
-				$a[12] = $a[15]; 
-				$a[13] = $a[16]; 
-				$a[14] = $a[17];
-			}
-
-
 			$label = "clinical genotype to phenotype annotations for ".$a[1];
+			
 			// [0] => Clinical Annotation Id
 			parent::addRDF(
 				parent::describeIndividual($id, $label, parent::getVoc()."Clinical-Annotation").
@@ -879,17 +877,15 @@ function clinical_ann_metadata()
 			// [5] => Genotype-Phenotypes IDs
 			// [6] => Text
 			if($a[5]) {
-				$gps = explode('","',$a[5]);
-				$gps_texts = explode('","',$a[6]);
+				$gps = explode(';',$a[5]);
+				$gps_texts = explode('; ',$a[6]);
 				foreach($gps AS $i => $gp) {
-					$gp = str_replace('"','',trim($gp));
-					$gp_text = str_replace('"','',trim($gps_texts[$i]));
-					$b = explode(":",$gp_text,2);
+					$gp_text = str_replace('\\','',$gps_texts[$i]);
 
 					parent::addRDF(
 						parent::describeIndividual(parent::getNamespace().$gp, $gp_text, parent::getVoc()."Genotype-Phenotype-Association").
 						parent::triplify($id, parent::getVoc()."genotype_phenotype", parent::getNamespace().$gp).
-						parent::triplifyString(parent::getNamespace().$gp, parent::getVoc()."genotype", trim($b[0])).
+						parent::triplifyString(parent::getNamespace().$gp, parent::getVoc()."genotype", trim($gp)).
 						parent::describeClass(parent::getVoc()."Genotype-Phenotype-Association", "PharmGKB Genotype Phenotype Association").
 						parent::describeProperty(parent::getVoc()."genotype_phenotype", "Relationship between a PharmGKB entity and a Genotype Phenotype").
 						parent::describeProperty(parent::getVoc()."genotype", "Relationship between a PharmGKB Genotype Phenotype and a genotype")
@@ -900,11 +896,14 @@ function clinical_ann_metadata()
 			// [7] => Variant Annotations IDs
 			// [8] => Variant Annotations
 			if($a[7]) {
-				$b = explode('","',$a[7]);
-				$b_texts =  explode('","',$a[8]);
+				$b = explode(';',$a[7]);
+				$b_texts =  explode(';',$a[8]);
+				if(count($b) != count($b_texts)) {
+					trigger_error("Error in parsing variant annotations");
+					exit();
+				}
 				foreach($b AS $i => $variant) {
-					$variant = str_replace('"','',trim($variant));
-					$variant_text = str_replace('"','',trim ($b_texts[$i]));
+					$variant_text = trim($b_texts[$i]);
 					parent::addRDF(
 						parent::describeIndividual(parent::getNamespace().$variant, $variant_text, parent::getVoc()."Variant-Annotation").
 						parent::triplify($id, parent::getVoc()."variant", parent::getNamespace().$variant)
@@ -914,7 +913,7 @@ function clinical_ann_metadata()
 
 			// [9] => PMIDs
 			if($a[9]) {
-				$b = $this->parseList($a[9]);
+				$b = explode(';', $a[9]);
 				foreach($b AS $i => $pmid) {
 					parent::addRDF(
 						parent::triplify($id, parent::getVoc()."article", "pubmed:".$pmid)
@@ -930,10 +929,9 @@ function clinical_ann_metadata()
 				);
 			}
 
-			// [11] => Related Drugs
+			// [11] => Related Chemicals
 			if($a[11]) {
-				//print_r($a);exit;
-				$b = $this->parseList($a[11]);	
+				$b = explode(';', $a[11]);
 				foreach($b AS $drug_label) {
 					preg_match('/\(PA(.*)\)/',$drug_label,$m);
 
@@ -951,7 +949,7 @@ function clinical_ann_metadata()
 			}
 			// [12] => Related Diseases
 			if($a[12]) {
-				$b = $this->parseList($a[12]);
+				$b = explode(';', $a[12]);
 				foreach($b AS $disease_label) {
 					preg_match('/\(PA(.*)\)/',$disease_label,$m);
 					if(isset($m[1])) {

Diff for: sgd/sgd.php

@@ -126,8 +126,8 @@ function process(){
 			 "complex"     => "curation/literature/go_protein_complex_slim.tab",
 			 "interaction" => "curation/literature/interaction_data.tab",
 			 "phenotype"   => "curation/literature/phenotype_data.tab",
-			 "pathways"    => "curation/literature/biochemical_pathways.tab",
-			 "mapping"     => "mapping"
+			 "pathways"    => "curation/literature/biochemical_pathways.tab"#,
+			 #"mapping"     => "mapping"
  		);
 
 		$graph_uri = parent::getGraphURI();
@@ -468,7 +468,7 @@ function features()
 					);
 				}
 				if($a[1] == "ORF" && $a[4] != '') {
-					$p2 = ucfirst(strtolower(str_replace(array("(",")"), array("%28","%29"), $a[4])))."p";
+					$p2 = ucfirst(strtolower(str_replace(array("(",")",","), array("%28","%29","%2C"), $a[4])))."p";
 					$p2label = "$p2";
 					$this->AddRDF(
 						parent::triplify($sid, $this->getVoc()."encodes", "sgd:$p2").
@@ -507,7 +507,7 @@ function features()
 
 			// common names
 			if($a[4]) {
-				$nid = str_replace(array("(",")"), array("%28","%29"), $a[4]);
+				$nid = str_replace(array("(",")",","), array("%28","%29","%2C"), $a[4]);
 				$this->AddRDF(
 					parent::triplifyString($sid, $this->getVoc()."standardName", $a[4]).
 					parent::triplify($sid, "owl:sameAs", "sgd:$nid").
@@ -526,7 +526,7 @@ function features()
 			// parent feature
 			$parent_type = '';
 			if($a[6]) {
-				$parent = str_replace(array("(",")"," "), array("%28","%29","_"), $a[6]);
+				$parent = str_replace(array("(",")"," ",","), array("%28","%29","_","%2C"), $a[6]);
 				$this->addRDF(
 					parent::triplify($sid, $this->getVoc()."is-proper-part-of", $this->getRes().$parent).
 					parent::describeProperty($this->getVoc()."is-proper-part-of", "Relationship between an SGD entity and an entity it is a proper part of")
@@ -1125,7 +1125,7 @@ function pathways(){
 
 			$eid = '';
 			if($a[3]) { // there is a protein
-				$eid = ucfirst(strtolower($a[3]))."p";
+				$eid = ucfirst(strtolower(str_replace(",","%2C",$a[3])))."p";
 				$this->AddRDF(
 					parent::triplify($this->getRes().$pid, $this->getVoc()."has-participant", $this->getRes().$eid)
 				);