def statAll(self):
out = open('%s/reads_stat.xls'%self.path['QC'],'w')
out.write('compact\tsample_name\tdata_type\tlib_method\traw_reads\tpandaseq_reads\tHQ_reads\tHQ_ratio\tTotal_ratio\tneeded_reads\tneed_to_reseq\n')
for compact,data_type_hash in self.sample_struct.iteritems():
for data_type,lib_method_hash in data_type_hash.iteritems():
for lib_method,sampleinfo in lib_method_hash.iteritems():
for sample,infos in sampleinfo.iteritems():
t = threading.Thread(target=self.stat_raw_reads,args=(compact,data_type,lib_method,sample))
self.__active_threads.add(t)
t.start()
while True:
if threading.activeCount() < self.concurrency:
break
for t in threading.enumerate():
if t in self.__active_threads:
t.join()
sort_sample_file = '%s/sam_barcode.all'%self.path['split']
for (compact,sample_name,barcode_info,data_type,lib_method,needed_reads) in parse_sam_all(sort_sample_file):
key_list = [compact,data_type,lib_method,sample_name]
if not self.check_keys( key_list , self.sample_struct ):
out.write('%s\t%s\t%s\t%s\t%s\t%s\n'%(compact,sample_name,data_type,lib_method,'None'))
continue
item = self.sample_struct[compact][data_type][lib_method][sample_name]
if int(item['raw_reads']) != 0:
t_ratio = int(item['HQ_reads']) / int(item['raw_reads']) * 100
else:
t_ratio = 0
out_str = str(MyList((item['raw_reads'],item['pandaseq_reads'],item['HQ_reads'],item['HQ_ratio'])))
need_to_reseq = int(needed_reads) - int(item['HQ_reads'])
out.write('%s\t%s\t%s\t%s\t%s\t%2.2f%%\t%s\t%s\n'%(compact,sample_name,data_type,lib_method,out_str,t_ratio,needed_reads,need_to_reseq))
out.close()
def get_info(self):
self.compact_data_type = {}
self.needed_reads = {}
sam_barcode_file = '%s/sam_barcode.all'%self.path['split']
for (compact,sample_name,barcode_info,data_type,lib_method,data_needed) in parse_sam_all(sam_barcode_file):
compact_path = '%s/%s'%(self.path['QC'],compact)
if compact not in self.compact_data_type:
self.compact_data_type[compact] = data_type
self.needed_reads[compact] = {}
elif self.compact_data_type[compact] != data_type:
stderr.write('The compact %s has two diffrent data_type!'%compact)
self.needed_reads[compact][sample_name] = int( data_needed )
def getSampleStruct(self):
self.sample_struct = {}
sam_file = '%s/sam_barcode.all'%self.path['split']
for ( compact,sample_name,barcode_info,data_type,lib_method,needed_reads ) in parse_sam_all(sam_file):
if compact not in self.sample_struct:
self.sample_struct[compact] = {}
if data_type not in self.sample_struct[compact]:
self.sample_struct[compact][data_type] = {}
if lib_method not in self.sample_struct[compact][data_type]:
self.sample_struct[compact][data_type][lib_method] = {}
if sample_name not in self.sample_struct[compact][data_type][lib_method]:
self.sample_struct[compact][data_type][lib_method][sample_name] = {
'pandaseq_reads' : 0,
'HQ_reads' : 0,
'HQ_ratio' : 0,
'raw_reads' : 0,
'needed_reads' : int(needed_reads),
}
if lib_method not in self.total_reads:
self.total_reads[lib_method] = 0
def _create_samples(self):
for sam_barcode_file in self.sam_barcode_files:
lib_method = get_lib_method(sam_barcode_file)
if lib_method == None:
continue
sys.stdout.write('sam_barcode_file: %s ... ok\n'%sam_barcode_file)
for (compact,sample_name,barcode_info,data_type,lib_method,needed_data) in parse_sam_all(sam_barcode_file):
sample = WorkPerSample(self.work_path,compact,sample_name,lib_method,data_type)
if not sample:
continue
yield sample
