dirJob = sys.argv[1]
N = int(sys.argv[2])
temp_age_folder = sys.argv[3]
temp_OS_folder = sys.argv[4]
from collections import defaultdict
# Running on rnuuspr66 server
# temporary folder to store qsub job related files
if not os.path.exists(dirJob):
os.mkdir(dirJob)
fusion = getallfusion()
fusion = list(fusion)
# randomly partition the fusion list into N nearly equal parts
fusionPartition = randompartition(fusion,N)
for i in range(N):
name = 'Job' + str(i) + '_datafile'
filename = os.path.join(dirJob,name)
ofile = open(filename,'w')
fp = fusionPartition[i]
for f in fp:
line = '\t'.join([f[0],f[1],f[2]])
ofile.write(line+'\n')
ofile.close()
for i in range(N):
name = 'Job' + str(i) + '_datafile'
filename = os.path.join(dirJob,name)
qsubString = '''#!/bin/sh
#PBS -N AgeOSJob%d
temp_OS_folder = sys.argv[4]
from collections import defaultdict
# Running on rnuuspr66 server
# temporary folder to store qsub job related files
if not os.path.exists(dirJob):
os.mkdir(dirJob)
fusion = getallfusion()
fusion = list(fusion)
# randomly partition the fusion list into N nearly equal parts
fDict = defaultdict(list)
for cancer,g5p,g3p in fusion:
fDict[(cancer,g3p)].append(g5p)
fusionPartition = randompartition(fDict.keys(),N)
for i in range(N):
name = 'Job' + str(i) + '_datafile'
filename = os.path.join(dirJob,name)
ofile = open(filename,'w')
fp = fusionPartition[i]
for f in fp:
g5p = fDict[f]
if len(g5p)==1:
gene5p = g5p[0]
else:
gene5p = ','.join(g5p)
line = '\t'.join([f[0],gene5p,f[1]])
ofile.write(line+'\n')
ofile.close()