Merge pull request #28 from IHEC/add_nome_seq

catch key error in validate semantics

version_metadata/json_spec/ihec_spec.json

@@ -35,6 +35,22 @@
         "CHIP_PROTOCOL_CROSSLINK_TIME",
         "LIBRARY_GENERATION_FRAGMENT_SIZE_RANGE"
     ],
+
+
+    "chip-seq_input": [
+        "EXPERIMENT_TYPE",
+        "EXPERIMENT_ONTOLOGY_URI",
+        "LIBRARY_STRATEGY",
+        "MOLECULE_ONTOLOGY_URI",
+        "MOLECULE",
+        "EXTRACTION_PROTOCOL",
+        "EXTRACTION_PROTOCOL_TYPE_OF_SONICATOR",
+        "EXTRACTION_PROTOCOL_SONICATION_CYCLES",
+        "CHIP_PROTOCOL",
+        "CHIP_PROTOCOL_CHROMATIN_AMOUNT"
+    ],
+
+
     "chromatin_accessibility": [
         "EXPERIMENT_TYPE",
         "EXPERIMENT_ONTOLOGY_URI",
@@ -265,5 +281,13 @@
         "LIBRARY_GENERATION_PCR_R_PRIMER_SEQUENCE",
         "LIBRARY_GENERATION_PCR_PRIMER_CONC",
         "LIBRARY_GENERATION_PCR_PRODUCT_ISOLATION_PROTOCOL"
-    ]
-}
+    ],
+
+"patches" : {
+	"RNA_PREPARATION_3'_RNA_ADAPTER_LIGATION_PROTOCOL" : "RNA_PREPARATION_3'_RNA ADAPTER_LIGATION_PROTOCOL",
+	"EXTRACTION_PROTOCOL_RNA_ENRICHMENT": "EXTRACTION_PROTOCOL_MRNA_ENRICHMENT",
+	"RNA_PREPARATION_FRAGMENT_SIZE_RANGE": "MRNA_PREPARATION_FRAGMENT_SIZE_RANGE",
+	"PREPARATION_INITIAL_RNA_QNTY": "PREPARATION_INITIAL_MRNA_QNTY"
+}
+
+}

version_metadata/review.py

@@ -1,23 +1,31 @@
 from utils import json2, cmn
 import config
 import sraparse
+import os 
 
 def objtype(e):
 	if 'BIOMATERIAL_TYPE' in e['attributes']:
 		return cmn.demanduniq(e['attributes']['BIOMATERIAL_TYPE']).lower().replace(' ', '_')
 	elif 'EXPERIMENT_TYPE' in e['attributes']:
-		return cmn.demanduniq(e['attributes']['EXPERIMENT_TYPE']).lower().replace(' ', '_')	
+		et = cmn.demanduniq(e['attributes']['EXPERIMENT_TYPE']).lower().replace(' ', '_')	
+		if et.startswith('histone'):
+			et = 'chip-seq'
+		elif et.find('input') != -1:
+			et = 'chip-seq_input'
+		return et
 	else:
 		return None
 def main(args):
 	#assert args.has('-sample')
 	xmls = args.args()
-	ihec = json2.loadf('./json_spec/ihec_spec.json')
+	schome = os.path.dirname(os.path.realpath(__file__))
+	ihec = json2.loadf('{0}/json_spec/ihec_spec.json'.format(schome))
 	if args.has('-isxml'):
 		loader = lambda x: sraparse.SRAParseObjSet.from_file(x).attributes()	
 	else:
 		loader = lambda x: json2.loadf(x)
 
+	patches = ihec["patches"]
 	for arg in xmls:
 		db = loader(arg)
 		for e in db:
@@ -26,7 +34,8 @@ def main(args):
 			required = ihec.get(ot, None)
 			if required:
 				missing = [x for x in required if not x in e['attributes']]
-				print alias, missing, ot
+				missing2 = [x for x in missing if not patches.get(x, '-')  in e['attributes']]
+				print alias, ot, missing2, 'XXXX', missing
 			else:
 				print alias, '__unknown_object_type__', ot
 

version_metadata/validate_experiment.py

@@ -42,14 +42,17 @@ def validate_semantics_stub(self, attrs):
 			return True
 
 	def validate_semantics(self, attrs):
-		attributes = attrs['attributes']
-		miRNA_experiment_type =  attributes['experiment_type'] in ['smRNA-Seq'] # abstract all this using a Rule interface... another day
-		miRNA_strategy = attributes['library_strategy'] in ['miRNA-Seq']
-		validation_status = miRNA_strategy 	if miRNA_experiment_type else not miRNA_strategy
-		if not validation_status:
-			logger.warn('#warn: __semantic_validation_failed__: smRNA-Seq library strategy if and only if miRNA-Seq experiment type\n')
-		return validation_status
-
+		try:
+			attributes = attrs['attributes']
+			miRNA_experiment_type =  attributes['experiment_type'] in ['smRNA-Seq'] # abstract all this using a Rule interface... another day
+			miRNA_strategy = attributes['library_strategy'] in ['miRNA-Seq']
+			validation_status = miRNA_strategy 	if miRNA_experiment_type else not miRNA_strategy
+			if not validation_status:
+				logger.warn('#warn: __semantic_validation_failed__: smRNA-Seq library strategy if and only if miRNA-Seq experiment type\n')
+			return validation_status
+		except KeyError as e:
+			logger.warn('#warn keyerror in validate_semantics, probably is not even syntactically valid\n')
+			return False
 
 # to do: refactor this code along with validate_sample into one module
 def main(args):