def excuteCreat(projdir,processes,samples):
structure_model = directory_structure.projdirectory(projdir,directory_structure.directory)
filtered_modelbyProcesses = final_path(structure_model,processes)
real_structure = creatDirect(filtered_modelbyProcesses,samples)
for adirname in real_structure:
for aprocess in real_structure[adirname]:
creatDir(real_structure[adirname][aprocess])
print '\t'.join([adirname,aprocess,real_structure[adirname][aprocess]])
dirs[dirname][aprocess] = re.sub(p,dirname,pathes[aprocess])
return dirs
def excuteCreat(projdir,processes,samples):
structure_model = directory_structure.projdirectory(projdir,directory_structure.directory)
filtered_modelbyProcesses = final_path(structure_model,processes)
real_structure = creatDirect(filtered_modelbyProcesses,samples)
for adirname in real_structure:
for aprocess in real_structure[adirname]:
creatDir(real_structure[adirname][aprocess])
print '\t'.join([adirname,aprocess,real_structure[adirname][aprocess]])
def write_shell(shell_file,cmds):
with open(shell_file,'w') as sh:
sh.write('\n'.join(cmds)+'\n')
if __name__ == "__main__":
p = directory_structure.projdirectory('d',directory_structure.directory)
for k in p:
print '\t'.join([k,p[k]])
print '#################################################################'
pp = final_path(p,['rawdata','cleandata','aln','somaticsnv','snv','somaticsv','somaticmsi'])
for k in pp:
print '\t'.join([k,pp[k]])
print '#################################################################'
dirs = creatDirect(pp,{'PA172':{'T':{'PA172TQ1':{}},'N':{'PA172NB1':{}}},'PA173':{'N':{'PA173NB1':{}}}})
for k in dirs:
for m in dirs[k]:
print '\t'.join([k,m,dirs[k][m]])
for asample in datas[aperson][TorN]:
for alib in datas[aperson][TorN][asample]:
for lane in datas[aperson][TorN][asample][alib]:
if datas[aperson][TorN][asample][alib][lane]['rawdata'] != '-':
if '.raw.fq.gz' not in infiles:
infiles['.raw.fq.gz'] = [datas[aperson][TorN][asample][alib][lane]['rawdata'].split(',')]
else:
infiles['.raw.fq.gz'].append(datas[aperson][TorN][asample][alib][lane]['rawdata'].split(','))
elif datas[aperson][TorN][asample][alib][lane]['cleandata'] != '-':
if '.clean.fq.gz' not in infiles:
infiles['.clean.fq.gz'] = [datas[aperson][TorN][asample][alib][lane]['cleandata'].split(',')]
else:
infiles['.clean.fq.gz'].append(datas[aperson][TorN][asample][alib][lane]['cleandata'].split(','))
##make dirs, and check the completeness of result files that were exsited
p = projdirectory(projdir,directory)
# for k in p:
# print '\t'.join([k,p[k]])
# print '#################################################################'
processes = set([wanted_jobs[ajob]['outputdir'] for ajob in wanted_jobs])
pp = final_path(p,processes)
# for k in pp:
# print '\t'.join([k,pp[k]])
# print '#################################################################'
dirs = creatDirect(pp,samples)
# for k in dirs:
# for m in dirs[k]:
# print '\t'.join([k,m,dirs[k][m]])
#寻找需要更新的bam文件和需要添加到datas中的bam文件，如以前下机的T，现在下机了N，可以做后面的somatic检测了，那就需要添加
#需要更新的bam文件是指加测新lane