diff spring_minz.py @ 25:5d1ae615e4ec draft

"planemo upload commit ff7d7e512b79436c3538078552983762330a920d-dirty"

--- a/spring_minz.py	Thu Oct 29 13:09:57 2020 +0000
+++ b/spring_minz.py	Sat Oct 31 22:01:29 2020 +0000
@@ -41,7 +41,6 @@
                     inputPath=inputPath,
                     crossReference=crossReference,
                     minScore=args.minscore,
-                    idLength=args.idx,
                     logFile=logFile,
                     interactions=interactions)
     if args.inputlist:
@@ -54,7 +53,6 @@
                         inputPath=targetPath,
                         crossReference=crossReference,
                         minScore=args.minscore,
-                        idLength=args.idx,
                         logFile=logFile,
                         interactions=interactions)
     interactions = sorted(interactions.values(), key=lambda item: item["minZ"], reverse=True)
@@ -63,15 +61,15 @@
             output_file.write("%s\t%s\t%s\t%s\n" % (entry["targetName"], entry["inputName"], entry["minZ"], entry["minInfo"]))
     logFile.close()
 
-def matchScores(targetFile, targetName, inputs, inputPath, crossReference, minScore, idLength, logFile, interactions):
-    targetTop, targetHits = getTemplateScores(targetFile, minScore, idLength)
+def matchScores(targetFile, targetName, inputs, inputPath, crossReference, minScore, logFile, interactions):
+    targetTop, targetHits = getTemplateScores(targetFile, minScore)
     if not targetHits:
         print("No targets found `%s`" % targetFile)
     else:
         print ("Loaded target scores from `%s`." % targetFile)
         for inputName in inputs:
             inputFile = "%s/%s" % (inputPath, inputName)
-            inputTop, inputHits = getTemplateScores(inputFile, minScore, idLength)
+            inputTop, inputHits = getTemplateScores(inputFile, minScore)
             minZ = 0
             minInfo = ""
             for t in targetHits:
@@ -94,17 +92,16 @@
                 interactions[interactionKey] = dict(targetName=targetName, inputName=inputName, minZ=minZ, minInfo=minInfo)
                 logFile.write("Interaction between %s and %s [min-Z: %s].\n" % (targetName, inputName, minZ))
 
-def getTemplateScores(hhrFile, minScore, identifierLength):
+def getTemplateScores(hhrFile, minScore):
     result = dict()
     topTemplate = None
-    identifierLength = identifierLength + 4
     if os.path.isfile(hhrFile):
         with open(hhrFile) as file:
             for index, line in enumerate(file):
                 if index > 8:
                     if not line.strip():
                         break
-                    templateId = line[4:identifierLength]
+                    templateId = line[4:10]
                     templateScore = float(line[57:63])
                     if templateScore > minScore:
                         if topTemplate is None:
@@ -119,7 +116,6 @@
     parser.add_argument('-il', '--inputlist', help='Text file containing identifiers.', required=False)
     parser.add_argument('-ip', '--inputpath', help='Directory containing `hhr` files', required=False)
     parser.add_argument('-c', '--crossreference', help='Cross Reference index file', required=True)
-    parser.add_argument('-x', '--idx', help='Length of identifier', type=int, default=6)
     parser.add_argument('-o', '--output', help='Output file containing min-Z scores', required=True)
     parser.add_argument('-l', '--log', help='Log file', required=True)
     parser.add_argument('-m', '--minscore', help='min-Z score threshold', type=int, default=10)