Merge pull request #463 from micheldumontier/release3

fixes for various parsers

Author: micheldumontier

drugbank/drugbank.php

@@ -36,7 +36,7 @@ class DrugBankParser extends Bio2RDFizer
     function __construct($argv) {
 		parent::__construct($argv,"drugbank");
 		parent::addParameter('files', true, 'all|drugbank','all','Files to convert');
-		parent::addParameter('download_url',false,null,'http://www.drugbank.ca/system/downloads/current/');
+		parent::addParameter('download_url',false,null,'https://www.drugbank.ca/releases/5-0-5/downloads/all-full-database');
 		parent::initialize();
     }
 
@@ -70,8 +70,9 @@ function Run()
 		$dataset_description = '';
 		foreach($files AS $f) {
 			if($f == 'drugbank') {
-				$file = 'drugbank.xml.zip';
+				$file = 'drugbank_all_full_database.xml.zip';
 				$lname = 'drugbank';
+				$insidezip_file = "full database.xml";
 			}
 			$fnx = 'parse_'.$f;
 
@@ -91,7 +92,7 @@ function Run()
 			if(file_exists($indir.$file)) {
 				// call the parser
 				echo "processing $file ...".PHP_EOL;
-				$this->$fnx($indir,$file);
+				$this->$fnx($indir,$file, $insidezip_file);
 				echo "done".PHP_EOL;
 				parent::clear();
 			}
@@ -146,9 +147,9 @@ function Run()
     }
 
 
-	function parse_drugbank($ldir,$infile)
+	function parse_drugbank($ldir,$infile, $insidezip_file)
 	{
-		$xml = new CXML($ldir.$infile);
+		$xml = new CXML($ldir.$infile, $insidezip_file);
 		while($xml->parse("drug") == TRUE) {
 			if(isset($this->id_list) and count($this->id_list) == 0) break;
 			$this->parseDrugEntry($xml);

goa/goa.php

@@ -57,22 +57,22 @@ function run(){
 
 		foreach($files as $file){
 			$download = parent::getParameterValue('download');
-			$lfile = $ldir."goa_".$file.".gz";
+			$basefile = "goa_".$file.".gaf";
+			$lfile = $ldir."/".$basefile.".gz";
 			if(!file_exists($lfile) && $download == false) {
 				trigger_error($lfile." not found. Will attempt to download.", E_USER_NOTICE);
 				$download = true;
 			}
 
 			//download file
-			$rfile = $rdir.strtoupper($file)."/gene_association.goa_".$file.".gz";
+			$rfile = $rdir.strtoupper($file)."/".$basefile.".gz";
 			if($download == true) {
-				echo "downloading $file ... ";
-				//file_put_contents($lfile,file_get_contents($rfile));
+				echo "downloading $file".".gz ... ";
 				utils::DownloadSingle($rfile,$lfile);
 			}
 
 			$gz = (strstr(parent::getParameterValue('output_format'),".gz") === FALSE)?false:true;
-			$ofile = "goa_".$file.".".parent::getParameterValue('output_format'); 	
+			$ofile = $basefile.".".parent::getParameterValue('output_format'); 	
 
 			parent::setReadFile($lfile, TRUE);
 			parent::setWriteFile($odir.$ofile, $gz);
@@ -153,6 +153,9 @@ function process($file){
 			$goid = substr($fields[4],3);
 			$refs = $this->getDbReferences($fields[5]);
 			$eco = $this->getEvidenceCodeLabelArr($fields[6]);
+			if($eco === null) {
+				print_r($fields[6]);exit;
+			}
 			$aspect = $this->getAspect($fields[8]);
 			$label = $fields[9];
 			$synonyms =  explode("|", $fields[10]);
@@ -291,6 +294,15 @@ function getEvidenceCodeLabelArr($aec){
 				"IGI"=> array("Inferred from Genetic Interaction","0000316"),
 				"IEP"=> array("Inferred from Expression Pattern", "0000008")
 				);
+				
+			$htp = array(			
+				"HTP" => array("Inferred from High Throughput Experiment",""),
+				"HDA" => array("Inferred from High Throughput Direct Assay",""),
+				"HMP" => array("Inferred from Hight Throughput Mutant Phenotype",""),
+				"HGI" => array("Inferred from High Throughput Genetic Interaction",""),
+				"HEP" => array("Inferred from High Throughput Expression Pattern","")
+			);
+			
 			//computational analysis codes
 			$cac = array(
 				"ISS"=> array("Inferred from Sequence or Structural Similarity","0000027"),
@@ -319,9 +331,11 @@ function getEvidenceCodeLabelArr($aec){
 			$aac = array(
 				"IEA"=>array("Inferred from Electronic Annotation", "0000203")
 			);
-
+			
 			if(array_key_exists($aec, $ec)){
 				return array("experimental evidence code"=>$ec[$aec]);
+			}elseif(array_key_exists($aec, $htp)){
+				return array("high throughput code"=>$htp[$aec]);
 			}elseif(array_key_exists($aec, $cac)){
 				return array("computational analysis code"=>$cac[$aec]);
 			}elseif(array_key_exists($aec, $asc)){
@@ -330,10 +344,8 @@ function getEvidenceCodeLabelArr($aec){
 				return array("curator statement code"=>$csc[$aec]);
 			}elseif(array_key_exists($aec, $aac)){
 				return array("automatically assigned code"=>$aac[$aec]);
-			}elseif(array_key_exists($aec, $oec)){
-				return array("obsolete evidence code"=>$oec[$aec]);
 			}else{
-				return null;
+				return array("unmapped evidence code"=> $aec);
 			}
 		} else {
 			return null;

hgnc/hgnc.php

@@ -117,10 +117,12 @@ function Run(){
 	function process(){
 		$header = $this->getReadFile()->read(200000);
 		$header_arr = explode("\t", $header);
+
 		$h = array_flip($header_arr);
 
 		$c = count($h);
 		$n = 52;
+
 		if ($c != $n)
 		{
 			echo PHP_EOL;

kegg/kegg.php

@@ -675,6 +675,7 @@ function parseEntry($lfile)
 			}
 			if(in_array($k, array("INTERACTION","METABOLISM","TARGET"))) {
 				// dopamine D2-receptor antagonist [HSA:1813] [KO:K04145]
+				// K04348 K06268 K17610 K17611
 				$id = parent::getRes().md5($uri.$v);
 				$type = ucfirst(strtolower($k));
 				if(in_array($k, array("INTERACTION","METABOLISM"))) {
@@ -697,10 +698,14 @@ function parseEntry($lfile)
 				if(isset($m[1]) and !empty($m[1])) {					
 					foreach($m[1] AS $item) {			
 						$a = explode(':',$item);  // get the namespace
+						if(!isset($a[1])) {continue;} // skip this.
 						$b = explode(' ',$a[1]);
 						foreach($b AS $c) {
-							if(!strstr($item,"KO")) $i = "kegg:".$a[0].'_'.$c;
-							else $i = "kegg:".$c;
+							if(!strstr($item,"KO")) {
+								$i = "kegg:".$a[0].'_'.$c;
+							} else {
+								$i = "kegg:".$c;
+							}
 							parent::addRDF(
 								parent::triplify($id,parent::getVoc()."link",$i)
 							);
@@ -845,4 +850,4 @@ function parseKGML($lfile)
 }
 
 
-	
+	

miriam/miriam.php

@@ -175,14 +175,16 @@ function parseItem($item)
 				$rid = parent::getRes().str_replace(":","",$id)."_".($i+1);
 				$a = $myitem['@attributes'];
 				$rid_type = parent::getVoc().'restriction_type_'.$a['type'];
+				$page = '';
+				if(isset($myitem['link']) and strstr($myitem['link'],"http") !== FALSE) $page = $myitem['link'];
 
 				parent::addRDF(
 					parent::describeIndividual($rid, $a['desc'], parent::getVoc()."Restriction").
 					parent::describeClass(parent::getVoc()."Restriction", "Resource Restriction").
 					parent::triplify($rid, "rdf:type", $rid_type).
 					parent::describeClass($rid_type, $a['desc'], parent::getVoc()."Restriction").
 					parent::triplifyString($rid, "dct:description", $myitem['statement']).
-					parent::triplify($rid, "foaf:page", isset($myitem['link'])?$myitem['link']:"").
+					parent::triplify($rid, "foaf:page", $page).
 					parent::triplify($id, parent::getVoc()."restriction", $rid)
 				);
 		}}

refseq/refseq.php

@@ -212,7 +212,7 @@ function process(){
 						$p = "/(\S+)\=(.*)/";
 						preg_match($p, $aL, $m);
 						if(count($m)){
-							if($m[1] == "db_xref"){
+							if($m[1] == "db_xref" and strstr($m[2],"http") !== FALSE){
 								parent::AddRDF(
 									parent::triplify($feat_res, "rdfs:seeAlso", str_replace("\"", "", $m[2]))
 								);