def _write_match_tmp(signatures: dict, u2variants: dict, p_proteins: str,
p_uniprot2matches: str, start: str, stop: Optional[str],
output: str):
proteins = Store(p_proteins)
u2matches = Store(p_uniprot2matches)
with open(output, "wt", encoding="utf-8") as fh:
doc = getDOMImplementation().createDocument(None, None, None)
for uniprot_acc, protein in proteins.range(start, stop):
elem = doc.createElement("protein")
elem.setAttribute("id", uniprot_acc)
elem.setAttribute("name", protein["identifier"])
elem.setAttribute("length", str(protein["length"]))
elem.setAttribute("crc64", protein["crc64"])
try:
protein_entries = u2matches[uniprot_acc]
except KeyError:
pass
else:
for signature_acc in sorted(protein_entries):
try:
signature = signatures[signature_acc]
except KeyError:
# InterPro entry
continue
elem.appendChild(
_create_match(doc, signature,
protein_entries[signature_acc]))
finally:
elem.writexml(fh, addindent=" ", newl="\n")
protein_variants = u2variants.get(uniprot_acc, [])
for variant, length, crc64, matches in protein_variants:
elem = doc.createElement("protein")
elem.setAttribute("id", variant)
elem.setAttribute("name", variant)
elem.setAttribute("length", str(length))
elem.setAttribute("crc64", crc64)
for signature_acc in sorted(matches):
try:
signature = signatures[signature_acc]
except KeyError:
# InterPro entry
continue
elem.appendChild(
_create_match(doc, signature, matches[signature_acc]))
elem.writexml(fh, addindent=" ", newl="\n")
def _write_feature_tmp(features: dict, p_proteins: str,
p_uniprot2features: str, start: str,
stop: Optional[str], output: str):
proteins = Store(p_proteins)
u2features = Store(p_uniprot2features)
with open(output, "wt", encoding="utf-8") as fh:
doc = getDOMImplementation().createDocument(None, None, None)
# for uniprot_acc, protein in proteins.range(start, stop):
for uniprot_acc, protein_features in u2features.range(start, stop):
protein = proteins[uniprot_acc]
elem = doc.createElement("protein")
elem.setAttribute("id", uniprot_acc)
elem.setAttribute("name", protein["identifier"])
elem.setAttribute("length", str(protein["length"]))
elem.setAttribute("crc64", protein["crc64"])
for feature_acc in sorted(protein_features):
feature = features[feature_acc]
feature_match = protein_features[feature_acc]
match = doc.createElement("match")
match.setAttribute("id", feature_acc)
match.setAttribute("name", feature["name"])
match.setAttribute("dbname", feature["database"])
match.setAttribute("status", 'T')
match.setAttribute("model", feature_acc)
match.setAttribute("evd", feature["evidence"])
for loc in sorted(feature_match["locations"]):
# there is only one fragment per location
pos_start, pos_end, seq_feature = loc
lcn = doc.createElement("lcn")
lcn.setAttribute("start", str(pos_start))
lcn.setAttribute("end", str(pos_end))
if seq_feature:
lcn.setAttribute("sequence-feature", seq_feature)
match.appendChild(lcn)
elem.appendChild(match)
elem.writexml(fh, addindent=" ", newl="\n")