added support for HiSeq and NextSeq runs

SVN: 33189

deep_sequencing_unit/sourceTest/core-plugins/illumina-qgf/1/dss/drop-boxes/register-undetermined/register-undetermined.py

@@ -24,34 +24,40 @@ import shutil
 from ch.systemsx.cisd.openbis.generic.shared.api.v1.dto import SearchCriteria
 from ch.systemsx.cisd.openbis.generic.shared.api.v1.dto import SearchSubCriteria
 
-def renameFiles(dir, flowcellName):
-  print dir 
-  print flowcellName
-  for root, dirs, files in os.walk(dir):
-    for file in files:
-      print root + file
-      os.rename(root + '/' + file, root + "/" + flowcellName + "_" + file)
 
+def splitFolderName (folderName):
+    runFolder, lane = folderName.rsplit("-", 1)
+    
+    # check if it is a NextSeq run, as we use only the Flowcell code as a Sample code 
+    if ("NS" in runFolder):
+        return runFolder.split("_")[-1][1:] + ":" + lane
+        
+    return runFolder + ":" + lane
+
+def searchSample (transaction, sampleCode):
+    # Get the search service
+    search_service = transaction.getSearchService()
+
+    print("Searching for " + sampleCode)
+    # Search for the sample
+    sc = SearchCriteria()
+    sc.addMatchClause(SearchCriteria.MatchClause.createAttributeMatch(SearchCriteria.MatchClauseAttribute.CODE, sampleCode))
+    foundSamples = search_service.searchForSamples(sc)
+    assert(len(foundSamples), 1)
+    return foundSamples[0]
 
 def process(transaction):
 
-  # 140724_M01761_0084_000000000-A8TNL/
-
-  incomingPath = transaction.getIncoming().getPath()
-  incomingFolder = transaction.getIncoming().getName()
-  flowLane = incomingFolder + ":1"
-
-  # Get the search service
-  search_service = transaction.getSearchService()
-
-  # Search for the sample
-  sc = SearchCriteria()
-  sc.addMatchClause(SearchCriteria.MatchClause.createAttributeMatch(SearchCriteria.MatchClauseAttribute.CODE, flowLane))
-  foundSamples = search_service.searchForSamples(sc)
+    # expected incoming folder names:
+    # MiSeq: 141217_D00535_0040_BC6GHLANXX-1
+    # NextSeq: 141217_NS500318_0034_AH1766BGXX-1
+    # HiSeq: 141204_D00535_0035_BC5LPVANXX-8
 
-  if foundSamples.size() > 0:
+    incomingFolder = transaction.getIncoming().getName()
+    flowLaneName = splitFolderName(incomingFolder) 
+    flowLane = searchSample(transaction, flowLaneName)
 
     dataSet = transaction.createNewDataSet("UNDETERMINED_READS_DISTRIBUTION")
     dataSet.setMeasuredData(False)
-    dataSet.setSample(foundSamples[0])
-    transaction.moveFile(incomingPath, dataSet)
+    dataSet.setSample(flowLane)
+    transaction.moveFile(transaction.getIncoming().getPath(), dataSet)