def generate_motif_features(feature: CDSFeature,
motifs: List[HMMResult]) -> List[CDSMotif]:
""" Convert a list of HMMResult to a list of CDSMotif features """
# use a locus tag if one exists
locus_tag = feature.get_name()
if feature.locus_tag:
locus_tag = feature.locus_tag
motif_features = []
for i, motif in enumerate(motifs):
i += 1 # user facing, so 1-indexed
loc = feature.get_sub_location_from_protein_coordinates(
motif.query_start, motif.query_end)
prot_loc = FeatureLocation(motif.query_start, motif.query_end)
new_motif = CDSMotif(loc,
feature.get_name(),
prot_loc,
tool="nrps_pks_domains")
new_motif.label = motif.hit_id
new_motif.domain_id = 'nrpspksmotif_{}_{:04d}'.format(locus_tag, i)
new_motif.evalue = motif.evalue
new_motif.score = motif.bitscore
new_motif.detection = "hmmscan"
new_motif.database = "abmotifs"
new_motif.locus_tag = locus_tag
new_motif.translation = feature.translation[motif.query_start:motif.
query_end]
motif_features.append(new_motif)
return motif_features
def generate_domain_features(gene: CDSFeature, domains: List[HMMResult]) -> Dict[HMMResult, AntismashDomain]:
""" Generates AntismashDomain features for each provided HMMResult
Arguments:
gene: the CDSFeature the domains were found in
domains: a list of HMMResults found in the CDSFeature
Returns:
a dictionary mapping the HMMResult used to the matching AntismashDomain
"""
new_features = {}
domain_counts = defaultdict(int) # type: Dict[str, int]
for domain in domains:
loc = gene.get_sub_location_from_protein_coordinates(domain.query_start, domain.query_end)
# set up new feature
new_feature = AntismashDomain(loc, tool="nrps_pks_domains")
new_feature.domain = domain.hit_id
new_feature.locus_tag = gene.locus_tag or gene.get_name()
new_feature.detection = "hmmscan"
new_feature.database = "nrpspksdomains.hmm"
new_feature.evalue = domain.evalue
new_feature.score = domain.bitscore
new_feature.translation = gene.translation[domain.query_start:domain.query_end + 1]
domain_counts[domain.hit_id] += 1 # 1-indexed, so increment before use
domain_name = "{}_{}.{}".format(gene.get_name(), domain.hit_id, domain_counts[domain.hit_id])
new_feature.domain_id = "nrpspksdomains_" + domain_name
new_feature.label = domain_name
new_features[domain] = new_feature
return new_features
def __init__(self, feature: CDSFeature) -> None:
super().__init__(["id", "sequence", "domains", "modules"])
self.sequence = feature.translation
self.id = feature.get_name()
self.domains = [] # type: List[JSONDomain]
self.modules = [] # type: List[JSONModule]
def __init__(self, feature: CDSFeature) -> None:
super().__init__(['id', 'sequence', 'domains'])
self.sequence = feature.translation
self.id = feature.get_name()
self.domains = [] # type: List[JSONDomain]
