261b8674b99bfd1007e110b06c70973e6b347f67
mmaddren
  Mon Jul 11 13:31:45 2011 -0700
cleaner version of mkGeoPkg
diff --git python/programs/mkGeoPkg/mkGeoPkg python/programs/mkGeoPkg/mkGeoPkg
index 53a1d40..23a3e84 100755
--- python/programs/mkGeoPkg/mkGeoPkg
+++ python/programs/mkGeoPkg/mkGeoPkg
@@ -318,102 +318,84 @@
 				sample['!Sample_instrument_model'] = instrumentModels[stanza['seqPlatform']]
 			
 		sample['!Sample_data_processing'] = compositeUrl
 
 		if idNum in geoMapping and geoMapping[idNum] != 'Inconsistent':
 			sample['!Sample_geo_accession'] = geoMapping[idNum]
 		
 		softfile[firstStanza['metaObject']] = sample
 		
 	return softfile, fileList
 		
 		
 def createMicroArraySoftFile(database, composite, organism, compositeUrl, mdb, cv, track, md5sums, expIds, expVars, geoMapping, series, datatype, copyDirectory):
 	pass
 	
-	
-def copyFiles(fileList, downloadsDirectory, copyDirectory):
-
-	print 'Copying files:'
-	for filename in fileList:
-		print 'Copying file ' + filename + ' (' + count + '/' + len(fileList) + ' ...'
-		shutil.copy2(downloadsDirectory + filename, copyDirectory)
-		
 def main():
 	database = sys.argv[1]
 	composite = sys.argv[2]
 	submitStart = sys.argv[3]
 	submitSize = int(sys.argv[4])
 	organism = organisms[database]
 
 	mdbPath = '/cluster/home/mmaddren/kent/src/hg/makeDb/trackDb/' + organism + '/' + database + '/metaDb/alpha/' + composite + '.ra' #CHANGE
 	cvPath = '/cluster/home/mmaddren/kent/src/hg/makeDb/trackDb/cv/alpha/cv.ra' #CHANGE
 	trackPath = '/cluster/home/mmaddren/kent/src/hg/makeDb/trackDb/' + organism + '/' + database + '/' + composite + '.ra'
 	md5path = '/hive/groups/encode/dcc/analysis/ftp/pipeline/' + database + '/' + composite + '/md5sum.txt'
 
 	downloadsDirectory = '/hive/groups/encode/dcc/analysis/ftp/pipeline/' + database + '/' + composite + '/'
 	copyDirectory = '/cluster/home/mmaddren/kent/python/ucscgenomics/mkGeoPkg/' + composite
 	
 	compositeUrl = 'http://genome.ucsc.edu/cgi-bin/hgTrackUi?db=' + database + '&g=' + composite
 	
 	mdb = RaFile(mdbPath)
 	cv = CvFile(cvPath)
 	track = RaFile(trackPath)
 	md5sums = readMd5sums(md5path)
 	expIds, expVars, geoMapping, series, datatype = createMappings(mdb)
 	
 	submission = dict()
 	sortedIds = expIds.keys()
 	sortedIds.sort()
 	sortedIds = sortedIds[sortedIds.index(submitStart):sortedIds.index(submitStart) + submitSize]
-	print 'Generating soft using expIds ' + min(sortedIds) + ' to ' + max(sortedIds)
+	minId = min(sortedIds)
+	maxId = max(sortedIds)
+	print 'Generating soft using expIds ' + minId + ' to ' + maxId
 	for expId in sortedIds:
 		submission[expId] = expIds[expId]
 
 	if datatype.soft == HighThroughputSoftFile:
 		softfile, fileList = createHighThroughputSoftFile(database, composite, organism, compositeUrl, mdb, cv, track, md5sums, submission, expVars, geoMapping, series, datatype, copyDirectory)
 	elif datatype.soft == MicroArraySoftFile:
 		softfile, fileList = createMicroArraySoftFile(database, composite, organism, compositeUrl, mdb, cv, track, md5sums, submission, expVars, geoMapping, series, datatype, copyDirectory)
 	else:
 		raise Error('unsupported type')
 		
 	print 'Writing soft file'
 	outfileName = os.path.dirname(sys.argv[0]) + composite + '.soft'
 	outfile = open(outfileName, 'w')
 	outfile.write(str(softfile))
 	
-	#print 'Copying files'
-
 	fileString = outfileName
 	for file in fileList:
 		fileString = fileString + ' ' + downloadsDirectory + file
 		
-#callList = [ '/opt/aspera/connect/bin/ascp asp-geo@upload.ncbi.nlm.nih.gov:ENCODE/' + composite, '-i', '~/encode_geo_key/encode_geo_key.ppk', '-QTr', '-l300m' ]
-#callList.extend(fileList)
-#callList.append(')
-    
 	fileString.strip()
 	callString = '/opt/aspera/connect/bin/ascp -i ~/encode_geo_key/encode_geo_key.ppk -QTr -l300m ' + fileString + ' asp-geo@upload.ncbi.nlm.nih.gov:ENCODE/' + composite
-#subprocess.call(callList, shell=True)
-	outscript = open(composite + '_' + submitStart + '-' + str(int(submitStart) + submitSize) + '.sh', 'w')
+	outscript = open(composite + minId + '-' + maxId + '.sh', 'w')
 	outscript.write('#!/bin/sh\n\n')
 	outscript.write('/opt/aspera/connect/bin/ascp -i ~/encode_geo_key/encode_geo_key.ppk -QTr -l300m \\\n')
 	
 	outscript.write(os.path.dirname(sys.argv[0]) + composite + '.soft' + ' \\\n')
 	
 	for file in fileList:
 		outscript.write(downloadsDirectory + file + ' \\\n')
 		
 	outscript.write('asp-geo@upload.ncbi.nlm.nih.gov:ENCODE\n')
 	outscript.close()
-	#os.fchmod(composite + '_' + submitStart + '-' + str(int(submitStart) + submitSize) + '.sh', stat.S_IXUSR)
-	os.system('chmod +x ' + composite + '_' + submitStart + '-' + str(int(submitStart) + submitSize) + '.sh')
-#	outscript.write(callString)
-	#print callString
-	#os.system(callString)
 	
-	#copyFiles(fileList, downloadsDirectory, copyDirectory)
+	os.system('chmod +x ' + composite + minId + '-' + maxId + '.sh')
 		
 	print 'Finished!'
 	
 if __name__ == '__main__':
 	main()
\ No newline at end of file