def _compute_fdr_score(self, conf):
self.conf = conf
res_fun = CResultFunction(conf)
cand_file_list = res_fun.find_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)
target_num, decoy_num, all_num = 0, 0, len(psm_list)
for (title, pep, decoy, ranker) 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) 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
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 _rerank_result(self, conf):
res_fun = CResultFunction(conf)
result_file_list = res_fun.find_res_file(conf)
psm_list = []
for file_path in result_file_list:
psm_list += res_fun.load_psm(file_path)
feature_write_path = os.path.join(conf.INPUT_SVM_FOLDER, res_fun.get_taskid(conf))
print('aa', conf.INPUT_SVM_FOLDER, feature_write_path)
self._write_svm_file(psm_list, feature_write_path)
score_write_path = feature_write_path + ".txt"
self._svm_classify(feature_write_path, conf.MODEL_PATH, score_write_path)
#os.remove(feature_write_path)
score_list = self._get_svm_score(score_write_path)
assert(len(score_list) == len(psm_list))
self._update_psm_score(result_file_list, score_list)
def _load_result(self, path):
resFunc = CResultFunction(self.conf)
psm_cand_list = resFunc.load_psm(path)
psm_list = []
title_dict = {}
for (title, pep, decoy, ranker, delta_score) in psm_cand_list:
if title not in title_dict: title_dict[title] = []
title_dict[title].append((pep, decoy, ranker))
for title in title_dict:
for (pep, decoy, ranker) in title_dict[title]:
if ranker == 1:
if not decoy: # target
psm_list.append((title, pep, decoy, ranker))
else: # decoy, check the other peps with same score
cand_list = title_dict[title]
cur_decoy = True
for (pep2, decoy2, ranker2) in cand_list:
if pep2.score == pep.score and not decoy2:
cur_decoy = False
break
psm_list.append((title, pep, cur_decoy, ranker))
return psm_list
def _load_result(self, path):
resFunc = CResultFunction(self.conf)
psm_list = resFunc.load_single_psm(path)
return psm_list
def _write_fdr_result(self, conf, fdr_score):
resFunc = CResultFunction(conf)
taskid = resFunc.get_taskid(conf, False)
res_file_list = resFunc.find_res_file(conf)
res_plabel_file_list = resFunc.find_res_file(conf, label_flag = True)
pep_score, pep_line = {}, {}
t_p_line = {}
head_line = ""
title_score, title_line, title_plabel_line = {}, {}, {}
pep_plabel_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[3] + '@' + tmp[4] + '@' + tmp[6]
ac_list = tmp[11].split(';')
is_target = False
for ac in ac_list:
if not ac.startswith("REV_"):
is_target = True
break
if not is_target: continue
s = float(tmp[12])
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
for sq in pep_line:
line = pep_line[sq]
tmp = line.strip().split('\t')
title = tmp[0]
t_p_line[title] = 1
modification_lines = []
for file in res_plabel_file_list:
with open(file) as f:
lines = f.readlines()
for i in range(len(lines)):
if lines[i].startswith("name="):
title = lines[i].strip().split('=')[1].strip()
if title in title_line:
line = lines[i+1].strip()
title_plabel_line[title] = line
if title in t_p_line:
line = lines[i+1].strip()
pep_plabel_line[title] = line
elif lines[i].startswith("[Modification]"):
modification_lines = []
for j in range(i+1, len(lines)):
ind = lines[j].find('=')
if ind < 0: break
modification_lines.append(lines[j].strip())
sorted_res = sorted(title_score.items(), key=lambda d: d[1], reverse=True)
write_path = os.path.join(conf.result_folder, conf.XLINK_PSM + '-' + taskid + '.txt')
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')
write_path = os.path.join(conf.result_folder, conf.XLINK_PSM + '-' + taskid + '.plabel')
with open(write_path, 'w') as f:
f.write("[FilePath]\n")
f.write("File_Path=" + ";".join(conf.mgf_file_list) + "\n")
f.write("[Modification]\n")
for mod_line in modification_lines:
f.write(mod_line + '\n')
f.write("[xlink]\n")
f.write("xlink=UAA\n")
f.write("[Total]\n")
f.write("total=" + str(len(sorted_res)) + '\n')
spec_id = 0
for (r, s) in sorted_res:
spec_id += 1
one_line = title_plabel_line[r]
f.write("[Spectrum" + str(spec_id) + "]\n")
f.write("name=" + r + "\n")
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.XLINK_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')
write_path = os.path.join(conf.result_folder, conf.XLINK_PEP + '-' + taskid + '.plabel')
with open(write_path, 'w') as f:
f.write("[FilePath]\n")
f.write("File_Path=" + ";".join(conf.mgf_file_list) + "\n")
f.write("[Modification]\n")
for mod_line in modification_lines:
f.write(mod_line + '\n')
f.write("[xlink]\n")
f.write("xlink=UAA\n")
f.write("[Total]\n")
f.write("total=" + str(len(sorted_res)) + '\n')
spec_id = 0
for (r, s) in sorted_res:
spec_id += 1
one_line = pep_line[r]
tmp = one_line.strip().split('\t')
title = tmp[0]
f.write("[Spectrum" + str(spec_id) + "]\n")
f.write("name=" + title + "\n")
one_line = pep_plabel_line[title]
f.write(one_line + "\n")