def from_biopython(bio_feature: SeqFeature, feature: "CDSFeature" = None, # type: ignore
leftovers: Optional[Dict] = None, record: Any = None) -> "CDSFeature":
if leftovers is None:
leftovers = Feature.make_qualifiers_copy(bio_feature)
# grab mandatory qualifiers
transl_table = 1
if record:
transl_table = record.transl_table
if "transl_table" in leftovers:
transl_table = int(leftovers.pop("transl_table")[0])
# semi-optional qualifiers
protein_id = leftovers.pop("protein_id", [None])[0]
locus_tag = leftovers.pop("locus_tag", [None])[0]
gene = leftovers.pop("gene", [None])[0]
if not (gene or protein_id or locus_tag):
if "pseudo" in leftovers or "pseudogene" in leftovers:
gene = "pseudo%s_%s"
else:
gene = "cds%s_%s"
gene = gene % (bio_feature.location.start, bio_feature.location.end)
name = locus_tag or protein_id or gene
try:
_verify_location(bio_feature.location)
except Exception as err:
message = "invalid location for %s: %s" % (name, str(err))
raise SecmetInvalidInputError(message) from err
try:
translation = _ensure_valid_translation(leftovers.pop("translation", [""])[0],
bio_feature.location, transl_table, record)
except ValueError as err:
raise SecmetInvalidInputError(str(err) + ": %s" % name) from err
feature = CDSFeature(bio_feature.location, translation, gene=gene,
locus_tag=locus_tag, protein_id=protein_id,
translation_table=transl_table)
# grab optional qualifiers
feature.product = leftovers.pop("product", [""])[0]
sec_met = leftovers.pop("sec_met_domain", None)
if sec_met:
feature.sec_met = SecMetQualifier.from_biopython(sec_met)
gene_functions = leftovers.pop("gene_functions", [])
if gene_functions:
feature.gene_functions.add_from_qualifier(gene_functions)
feature.nrps_pks.add_from_qualifier(leftovers.pop("NRPS_PKS", []))
# grab parent optional qualifiers
super(CDSFeature, feature).from_biopython(bio_feature, feature=feature, leftovers=leftovers)
return feature
def from_biopython(bio_feature: SeqFeature, feature: "CDSFeature" = None, # type: ignore
leftovers: Optional[Dict] = None, record: Any = None) -> "CDSFeature":
if leftovers is None:
leftovers = Feature.make_qualifiers_copy(bio_feature)
# grab mandatory qualifiers
transl_table = 1
if record:
transl_table = record.transl_table
if "transl_table" in leftovers:
transl_table = int(leftovers.pop("transl_table")[0])
translation = leftovers.pop("translation", [""])[0]
# semi-optional qualifiers
protein_id = leftovers.pop("protein_id", [None])[0]
locus_tag = leftovers.pop("locus_tag", [None])[0]
gene = leftovers.pop("gene", [None])[0]
if not (gene or protein_id or locus_tag):
if "pseudo" in leftovers or "pseudogene" in leftovers:
gene = "pseudo%s_%s"
else:
gene = "cds%s_%s"
gene = gene % (bio_feature.location.start, bio_feature.location.end)
try:
_verify_location(bio_feature.location)
except Exception as err:
message = "invalid location for %s: %s" % (gene or protein_id or locus_tag, str(err))
raise SecmetInvalidInputError(message) from err
# ensure translation is valid if it exists
if translation:
invalid = set(translation) - _VALID_TRANSLATION_CHARS
if invalid:
logging.warning("Regenerating translation for CDS %s (at %s) containing invalid characters: %s",
locus_tag or protein_id or gene, bio_feature.location, invalid)
translation = ""
# ensure that the translation fits
if not _is_valid_translation_length(translation, bio_feature.location):
raise SecmetInvalidInputError("translation longer than location allows: %s > %s" % (
len(translation) * 3, len(bio_feature.location)))
# finally, generate the translation if it doesn't exist
if not translation:
if not record:
raise SecmetInvalidInputError("no translation in CDS and no record to generate it with")
if bio_feature.location.end > len(record.seq):
raise SecmetInvalidInputError("feature missing translation and sequence too short: %s" % (
(gene or protein_id or locus_tag)))
translation = record.get_aa_translation_from_location(bio_feature.location, transl_table)
assert _is_valid_translation_length(translation, bio_feature.location)
feature = CDSFeature(bio_feature.location, translation, gene=gene,
locus_tag=locus_tag, protein_id=protein_id,
translation_table=transl_table)
# grab optional qualifiers
feature.product = leftovers.pop("product", [""])[0]
sec_met = leftovers.pop("sec_met_domain", None)
if sec_met:
feature.sec_met = SecMetQualifier.from_biopython(sec_met)
gene_functions = leftovers.pop("gene_functions", [])
if gene_functions:
feature.gene_functions.add_from_qualifier(gene_functions)
# grab parent optional qualifiers
super(CDSFeature, feature).from_biopython(bio_feature, feature=feature, leftovers=leftovers)
return feature
