def compute_full_phmm(self, location):
"""realignes the fasta file stored in this object but to a different location"""
msa_string = self.generate_msa_string(self.fasta_hash.keys())
with tmp.NamedTemporaryFile() as msa_tmp:
write_to_tempfile(msa_tmp.name, msa_string)
generate_hmm(location, msa_tmp.name)
return location
def best_exonerate_prediction2(region_fasta, query_fasta, dir_path, hmm):
with tmp.NamedTemporaryFile() as reg_file:
write_to_tempfile(reg_file.name, region_fasta)
ex_obj = run_exonerate("-m p2g -E no", "{}.exon_p2g", dir_path,
region_fasta, query_fasta)
if ex_obj:
all_proteins = all_proteins_to_fasta_string(ex_obj)
TP_scores = markov_model_scoring2(all_proteins, hmm)
if TP_scores:
print(TP_scores)
max_score = max(list(TP_scores.values()))
max_score_header = set([
header.split(";")[0]
for header, score in TP_scores.items()
if score >= max_score * 0.90
])
fasta_hash = hash_fasta(query_fasta) # do that before
with tmp.NamedTemporaryFile() as q_file:
max_val_fasta = "\n".join([
"{}\n{}".format(header, fasta_hash[header])
for header in max_score_header
])
write_to_tempfile(q_file.name, max_val_fasta)
ex_name = "{}.exon".format(q_file.name)
ex_obj = run_exonerate("-m p2g -E no ", ex_name, dir_path,
region_fasta, q_file.name)
return ex_obj
return None
def markov_model_scoring2(fasta_string, hmm):
if hmm:
with tmp.NamedTemporaryFile() as ex_file:
write_to_tempfile(ex_file.name, fasta_string)
score_hash = get_phmm_score(hmm, ex_file.name)
if score_hash:
return score_hash
return None
def generate_msa_string(self, rest_prot):
"""returns a MSA as string in fasta format, derived from the a list of headers"""
seq_list = [
"{}\n{}".format(header, self.fasta_hash[header])
for header in rest_prot
]
with tmp.NamedTemporaryFile() as r_tmp:
write_to_tempfile(r_tmp.name, "\n".join(seq_list))
list_msa = generate_msa(r_tmp.name)
return "\n".join(list_msa)
def bulk_hmm_scoring_model():
with tempfile.NamedTemporaryFile() as tmp:
seq_list = []
for header in seq_header_list:
seq_list.append(header)
seq_list.append(cluster_hash[header])
write_to_tempfile(tmp.name, "\n".join(seq_list))
msa = "\n".join(generate_msa(tmp.name))
with tempfile.NamedTemporaryFile() as msa_tmp:
write_to_tempfile(msa_tmp.name, msa)
with tempfile.NamedTemporaryFile() as tmp_hmm:
hmm = generate_hmm(tmp_hmm.name, msa_tmp.name)
score_list = many_hmm_scores(hmm, tmp.name)
return score_list
def find_best_exonerate_result(region_tuple, region_fasta, group, cluster,
dir_path):
'''first aligns all proteins in p2g mode, scores them against the hmm and extracts the highest scoring predictions.
Then the corresponding actual proteins of those are re-aligned in p2g -E yes mode. Returned exonerate object,
holds thoses (few-best) predictions -> needs for further filtering.'''
query_fasta = data_base.group_by_cluster_to_fasta_file[group][cluster]
query_hash = data_base.group_by_cluster_to_fasta_hash[group][cluster]
hmm = data_base.group_by_cluster_to_hmm[group][cluster]
cutoff = data_base.group_by_cluster_to_score_cutoff[group][cluster]
length_range = data_base.group_by_cluster_to_length_range[group][cluster]
with tmp.NamedTemporaryFile() as reg_file:
write_to_tempfile(reg_file.name, region_fasta)
ex_obj = run_exonerate("-m p2g -E no", "{}.exon_p2g", dir_path,
reg_file.name, query_fasta)
try:
TP_scores = markov_model_scoring(
all_proteins_to_fasta_string(ex_obj), hmm)
if not quick:
max_score = max(list(TP_scores.values()))
max_header = set([
header.split(";")[0]
for header, score in TP_scores.items()
if score == max_score
])
with tmp.NamedTemporaryFile() as q_file:
max_val_fasta = "\n".join([
"{}\n{}".format(header, query_hash[header])
for header in max_header
])
write_to_tempfile(q_file.name, max_val_fasta)
ex_obj = run_exonerate("-m p2g -E yes",
"{}.exon".format(q_file.name),
dir_path, reg_file.name,
q_file.name)
TP_scores = markov_model_scoring(
all_proteins_to_fasta_string(ex_obj), hmm)
best_pred = []
for key_tuple in ex_obj.target_dna:
score_id = ">" + key_tuple.query + ";{}".format(key_tuple.idx)
pred_obj = PredictionObject(region_tuple, TP_scores[score_id],
cluster, cutoff, length_range)
pred_obj.infer_data_from_exonerate_obj(ex_obj, key_tuple)
best_pred.append(pred_obj)
filtered, passed = isolate_overlapping_predictions(
sorted(best_pred, key=lambda x: x.score, reverse=True))
return ex_obj, passed
except (AttributeError, TypeError, KeyError):
return None, None
def iterative_score_computation(self):
for idx in range(0, len(self.fasta_hash.keys())):
rest_prot = list(self.fasta_hash.keys())
query = rest_prot.pop(idx)
with tmp.NamedTemporaryFile() as q_tmp:
write_to_tempfile(
q_tmp.name,
query.split()[0] + "\n" + self.fasta_hash[query])
msa_string = self.generate_msa_string(rest_prot)
with tmp.NamedTemporaryFile() as msa_tmp:
write_to_tempfile(msa_tmp.name, msa_string)
with tmp.NamedTemporaryFile() as hmm_tmp:
generate_hmm(hmm_tmp.name, msa_tmp.name)
try:
score_dict = get_phmm_score(
hmm_tmp.name, q_tmp.name)
except IndexError:
continue
self.score_dict.update(score_dict)
return self.score_dict