SE-341 Improvements to handling the metabolomics data.

SVN: 21597

eu_basynthec/dist/etc/metabolomics/data-set-handler.py

 from datetime import datetime
 from eu.basynthec.cisd.dss import TimeSeriesDataExcel
 
-def retrieve_or_create_experiment(tr):
-	# Create an experiment
-	now_str = datetime.today().strftime('%Y%m%d-%H%M%S')
-	expid = "/SHARED/SHARED/" + now_str
-	exp = tr.getExperiment(expid)
+def retrieve_or_create_experiment(tr, exp_id):
+	"""Get the specified experiment form the server, or create a new one if no experiment is specified. Return the experiment."""
+	if exp_id is None:
+		now_str = datetime.today().strftime('%Y%m%d')
+		exp_id = "/SHARED/SHARED/" + now_str
+		exp = tr.getExperiment(exp_id)
+		if None == exp:
+			exp = tr.createNewExperiment(expid, 'BASYNTHEC')
+			exp.setPropertyValue("DESCRIPTION", "An experiment created on " + now_str)
+	else:
+		exp = tr.getExperiment(exp_id)
+	return exp
 
-	if None == exp:
-		exp = tr.createNewExperiment(expid, 'BASYNTHEC')
-		exp.setPropertyValue("DESCRIPTION", "An experiment created on " + now_str)
+def assign_properties(dataset, metadata):
+	"""Assign properties to the data set from information in the data."""
+	propertyNameMap = {
+		"STRAIN":"STRAIN", 
+		"TIMEPOINT TYPE": "TIMEPOINT_TYPE", 
+		"CELL LOCATION": "CELL_LOCATION", 
+		"VALUE TYPE": "VALUE_TYPE", 
+		"VALUE UNIT": "VALUE_UNIT", 
+		"SCALE": "SCALE"
+		}
 		
-	return exp
+	for prop in metadata.keySet():
+		key = propertyNameMap.get(prop)
+		if key is not None:
+			value = metadata.get(prop)
+			if (key == "STRAIN"):
+				value = value + " (STRAIN)"
+			dataset.setPropertyValue(key, value.upper())
+			
+def convert_data_to_tsv(tr, dataset, location):
+	"""Create a tsv file containing the data and add it to the data set."""
+	tr.createNewDirectory(dataset, location)
+	tsvFileName = tr.createNewFile(dataset, location, incoming.getName() + ".tsv")
+	tsv = open(tsvFileName, 'w')
+	for line in timeSeriesData.getRawDataLines():
+		for i in range(0, len(line) - 1):
+			tsv.write(line[i])
+			tsv.write("\t")
+		tsv.write(line[len(line) - 1])
+		tsv.write("\n")
+	tsv.close()
+	
+def store_original_data(tr, dataset, location):
+	"""Put the original data into the data set."""
+	tr.createNewDirectory(dataset, location)
+	tr.moveFile(incoming.getAbsolutePath(), dataset, location + "/" + incoming.getName())
 
-propertyNameMap = {
-	"STRAIN":"STRAIN", 
-	"TIMEPOINT TYPE": "TIMEPOINT_TYPE", 
-	"CELL LOCATION": "CELL_LOCATION", 
-	"VALUE TYPE": "VALUE_TYPE", 
-	"VALUE UNIT": "VALUE_UNIT", 
-	"SCALE": "SCALE"
-	}
 
 tr = service.transaction(incoming)
-exp = retrieve_or_create_experiment(tr)
-
 timeSeriesData = TimeSeriesDataExcel.createTimeSeriesDataExcel(incoming.getAbsolutePath())
 
 # create the data set and assign the metadata from the file
-dataSet = tr.createNewDataSet("METABOLITE_INTENSITIES")
+dataset = tr.createNewDataSet("METABOLITE_INTENSITIES")
 metadata = timeSeriesData.getMetadataMap()
+assign_properties(dataset, metadata)
+		
+# Convert the data into a tsv file, and put that and the original data into the data set
+convert_data_to_tsv(tr, dataset, "data/tsv")
+store_original_data(tr, dataset, "data/xls")
 
-for prop in metadata.keySet():
-	key = propertyNameMap.get(prop)
-	if key is not None:
-		value = metadata.get(prop)
-		if (key == "STRAIN"):
-			value = value + " (STRAIN)"
-		dataSet.setPropertyValue(key, value.upper())
-
-dataSet.setExperiment(exp)
-tr.moveFile(incoming.getAbsolutePath(), dataSet)
+expCode = metadata.get("EXPERIMENT")
+exp = retrieve_or_create_experiment(tr, expCode)
+dataset.setExperiment(exp)
 

eu_basynthec/source/java/eu/basynthec/cisd/dss/TimeSeriesDataExcel.java

@@ -133,7 +133,12 @@ public class TimeSeriesDataExcel
         for (int i = 1; i < metadataLines.size(); ++i)
         {
             String[] line = metadataLines.get(i);
-            metadataMap.put(line[0].toUpperCase(), line[1]);
+            String value = line[1];
+            if ("BLANK".equals(value))
+            {
+                value = null;
+            }
+            metadataMap.put(line[0].toUpperCase(), value);
         }
 
         return metadataMap;