def run_starter_unit_blastp(
cluster: Cluster,
cds_hmm_hits: Dict[str,
List[HMMResult]]) -> Dict[str, List[HMMResult]]:
""" Runs blastp on starter unit coding sequences in given cluster
Arguments:
cluster: Cluster on which the blastp shall be run
cds_hmm_hits: HMMResults by cds from type II PKS hmmscan
Returns:
None if no starter unit cds are present otherwise a dictionary of key: cds and value: list of HMMresults, for blastp results of the cluster
"""
starter_unit_cds = {}
for cds, hmm_hits in cds_hmm_hits.items():
starter_unit_hit_ids = [
hit.hit_id for hit in hmm_hits
if hit.hit_id in ['KSIII', 'AT', 'AMID', 'LIG']
]
if starter_unit_hit_ids:
starter_unit_cds[cluster.parent_record.get_cds_by_name(
cds)] = starter_unit_hit_ids
if starter_unit_cds:
blastp_results = []
blastp_fasta_files = set()
for cds, starter_unit_hit_ids in starter_unit_cds.items():
query_sequence = fasta.get_fasta_from_features([cds])
for hit_id in starter_unit_hit_ids:
blast_database = path.get_full_path(__file__, 'data', hit_id)
blastp_results.extend(
subprocessing.run_blastp(blast_database, query_sequence))
blastp_fasta_files.add(
path.get_full_path(__file__, 'data', hit_id + '.fasta'))
fasta_lengths = {}
for fasta_file in blastp_fasta_files:
fasta_lengths.update(get_fasta_lengths(fasta_file))
return refine_hmmscan_results(blastp_results, fasta_lengths)
return {}
def run_starter_unit_blastp(
cds_hmm_hits: Dict[CDSFeature,
List[HMMResult]]) -> Dict[str, List[HMMResult]]:
""" Runs blastp on starter unit coding sequences in given cluster
Arguments:
cds_hmm_hits: HMMResults by cds from type II PKS hmmscan
Returns:
a dictionary mapping CDS name to a list of HMMresults
"""
blastp_results = []
blastp_fasta_files = set()
for cds, hmm_hits in cds_hmm_hits.items():
query_sequence = fasta.get_fasta_from_features([cds])
for hit in hmm_hits:
if hit.hit_id not in ['KSIII', 'AT', 'AMID', 'LIG']:
continue
blast_database = path.get_full_path(__file__, 'data', hit.hit_id)
blastp_results.extend(
subprocessing.run_blastp(blast_database, query_sequence))
blastp_fasta_files.add(
path.get_full_path(__file__, 'data', hit.hit_id + '.fasta'))
if not blastp_results:
return {}
fasta_lengths = {}
for fasta_file in blastp_fasta_files:
fasta_lengths.update(get_fasta_lengths(fasta_file))
results = refine_hmmscan_results(blastp_results, fasta_lengths)
for hits in results.values():
for i, hit in enumerate(hits):
if not hit.hit_id.endswith("-CoA"):
hits[i] = HMMResult(hit.hit_id + "-CoA", hit.query_start,
hit.query_end, hit.evalue, hit.bitscore)
return results