def _write_fdr_result(self, conf, fdr_score):
resFunc = CResultFunction(conf)
taskid = resFunc.get_taskid(conf, False)
res_file_list = resFunc.find_single_res_file(conf)
pep_score, pep_line = {}, {}
head_line = ""
title_score, title_line = {}, {}
for file in res_file_list:
with open(file) as f:
lines = f.readlines()
head_line = lines[0].strip()
for i in range(1, len(lines)):
line = lines[i].strip()
tmp = line.split('\t')
title = tmp[0]
sq_key = tmp[2] + '@' + tmp[4] + '@' + str(
int(float(tmp[3]) + 0.5))
is_target = (tmp[8] == "True")
if not is_target: continue
s = float(tmp[5])
if s <= fdr_score: break
title_score[title] = s
title_line[title] = line
if sq_key not in pep_score or pep_score[sq_key] < s:
pep_score[sq_key] = s
pep_line[sq_key] = line
sorted_res = sorted(title_score.items(),
key=lambda d: d[1],
reverse=True)
write_path = os.path.join(conf.result_folder,
conf.SINGLE_PSM + '-' + taskid + '.txt')
conf.single_res_path = write_path
with open(write_path, 'w') as f:
f.write(head_line + '\n')
for (r, s) in sorted_res:
one_line = title_line[r]
f.write(one_line + '\n')
sorted_res = sorted(pep_score.items(),
key=lambda d: d[1],
reverse=True)
write_path = os.path.join(conf.result_folder,
conf.SINGLE_PEP + '-' + taskid + '.txt')
with open(write_path, 'w') as f:
f.write(head_line + '\n')
for (r, s) in sorted_res:
one_line = pep_line[r]
f.write(one_line + '\n')
def _compute_fdr_score(self):
conf = self.conf
res_fun = CResultFunction(conf)
cand_file_list = res_fun.find_single_res_file(conf)
print("[Info] Candidate file list", cand_file_list)
psm_list = []
for file_path in cand_file_list:
psm_list += self._load_result(file_path)
psm_list.sort(key=lambda k: k[1].score, reverse=True) #按照score排序
target_num, decoy_num, all_num = 0, 0, len(psm_list)
for (title, pep, decoy, ranker, delta_score) in psm_list:
if decoy:
decoy_num += 1
else:
target_num += 1
print("[Info] #PSMs is {0}, #Target is {1} and #Decoy is {2}".format(
all_num, target_num, decoy_num))
cur_target_num, cur_decoy_num = 0, 0
if len(psm_list) == 0: return 0.0
score_t = psm_list[0][1].score + 1.0
if target_num > 0 and decoy_num * 1.0 / target_num <= conf.fdr_value:
score_t = psm_list[-1][1].score
return score_t
for (title, pep, decoy, ranker, delta_score) in psm_list[::-1]:
if decoy:
cur_decoy_num += 1
else:
cur_target_num += 1
if target_num == cur_target_num:
fdr_value = 100
else:
fdr_value = (decoy_num - cur_decoy_num) * 1.0 / (
target_num - cur_target_num)
# print(decoy_num,cur_decoy_num, target_num, cur_target_num, fdr_value)
if fdr_value <= conf.fdr_value:
score_t = pep.score
return score_t
return score_t