def preprocess_cnv_patient_data(self, patients, is_pos=True):
# get the dictionary of gene id mappers
uni2ent, ent2uni = uniprot_mapper.json_to_dict()
res = []
for pat_id, patient in patients.items():
for cnv_type, ent_ids in patient.items():
patient[cnv_type] = [
uid for eid in ent_ids if eid in ent2uni
for uid in ent2uni[eid]
]
return res
def preprocess_seq_patient_data(self, GE, all_ent_ids):
# get the dictionary of gene id mappers
uni2ent, ent2uni = uniprot_mapper.json_to_dict()
found_ent_ids = [eid in ent2uni for eid in all_ent_ids]
ent_ids = np.array([eid for eid in all_ent_ids if eid in ent2uni])
uni_ids = np.array([ent2uni[eid] for eid in ent_ids], dtype=object)
log('uni_ids:', len(uni_ids))
log('miss_ent_ids:', len(all_ent_ids) - sum(found_ent_ids))
# prune genes whose uniprot id is not found
GE = GE[found_ent_ids]
return GE, uni_ids
def preprocess_som_patient_data(self, patients):
# get the dictionary of gene id mappers
uni2ent, ent2uni = uniprot_mapper.json_to_dict()
res = []
num_empty = 0
for pat_id, ent_ids in patients.items():
# uni_ids = [uid for eid in ent_ids if eid in ent2uni for uid in ent2uni[eid]]
uni_ids = [uid for eid in ent_ids if eid in ent2uni for uid in ent2uni[eid]]
# if there are any matches map them
res.append({
'pat_id': pat_id,
'mutated_nodes': uni_ids,
})
log('removed patients:', num_empty)
return res
def preprocess_patient_data(patients):
# get the dictionary of gene id mappers
uni2ent, ent2uni = uniprot_mapper.json_to_dict()
res = []
num_empty = 0
for pat_id, ent_ids in patients.items():
# uni_ids = [uid for eid in ent_ids if eid in ent2uni for uid in ent2uni[eid]]
uni_ids = [uid for eid in ent_ids if eid in ent2uni for uid in ent2uni[eid]]
# if there are any matches map them
if len(uni_ids) > 0:
res.append({
'pat_id': pat_id,
'mutated_nodes': uni_ids,
})
else:
num_empty += 1
return res