def run(login, gene_records, chrom_wdid):
for record in tqdm(gene_records):
if 'genomic_pos' not in record:
# see: http://mygene.info/v3/gene/855814
PBB_Core.WDItemEngine.log(
"WARNING",
format_msg(record['_id']['@value'], "no_position", '',
ENTREZ_PROP))
continue
if isinstance(record['genomic_pos']['@value'], list):
# see: http://mygene.info/v3/gene/853483
PBB_Core.WDItemEngine.log(
"WARNING",
format_msg(record['_id']['@value'], "multiple_positions", '',
ENTREZ_PROP))
continue
wd_item_construction(record, strain_info, chrom_wdid, login)
def run(login, records, add_pubmed):
# get all entrez gene id -> wdid mappings, where found in taxon is this strain
gene_wdid_mapping = WDHelper().id_mapper(
"P351", (("P703", strain_info['organism_wdid']), ))
# get all goID to wdid mappings
go_wdid_mapping = WDHelper().id_mapper("P686")
for record in tqdm(records, desc=strain_info['organism_name']):
entrez_gene = str(record['entrezgene']['@value'])
if entrez_gene not in gene_wdid_mapping:
PBB_Core.WDItemEngine.log(
"ERROR",
format_msg(record['_id']['@value'], "gene_not_found", None,
ENTREZ_PROP))
continue
gene_qid = gene_wdid_mapping[entrez_gene]
protein_item(record, strain_info, gene_qid, go_wdid_mapping, login,
add_pubmed)
def run_encodes(login, records):
# get all entrez gene id -> wdid mappings, where found in taxon is this strain
gene_wdid_mapping = PBB_Helpers.id_mapper(
"P351", (("P703", strain_info['organism_wdid']), ))
# get all ensembl protein id -> wdid mappings, where found in taxon is this strain
protein_wdid_mapping = PBB_Helpers.id_mapper(
"P705", (("P703", strain_info['organism_wdid']), ))
for record in tqdm(records, desc=strain_info['organism_name']):
entrez_gene = str(record['entrezgene']['@value'])
if entrez_gene not in gene_wdid_mapping:
PBB_Core.WDItemEngine.log(
"ERROR",
format_msg(record['_id']['@value'], "gene_not_found", None,
ENTREZ_PROP))
continue
gene_qid = gene_wdid_mapping[entrez_gene]
protein_qid = protein_wdid_mapping[record['ensembl']['@value']
['protein']]
gene_encodes_statement(gene_qid, protein_qid, 'ncbi_gene', entrez_gene,
record['ensembl']['@source'], login)
def protein_item(record, strain_info, gene_qid, go_wdid_mapping, login,
add_pubmed):
"""
generate pbb_core item object
"""
item_name = '{} {}'.format(record['name']['@value'],
record['ensembl']['@value']['protein'])
item_description = '{} protein found in {}'.format(
strain_info['organism_type'], strain_info['organism_name'])
s = []
############
# external IDs
############
# will be used for reference statements
external_ids = {
'entrez_gene': str(record['entrezgene']['@value']),
'ensembl_protein': record['ensembl']['@value']['protein'],
'ensembl_gene': record['ensembl']['@value']['gene'],
'refseq_protein': record['refseq']['@value']['protein'],
'uniprot': record['uniprot']['@value']['Swiss-Prot']
}
# ensembl protein id
ensembl_ref = make_ref_source(record['ensembl']['@source'],
'ensembl_protein',
external_ids['ensembl_protein'])
s.append(
PBB_Core.WDString(external_ids['ensembl_protein'],
'P705',
references=[ensembl_ref]))
# refseq protein id
refseq_ref = make_ref_source(record['refseq']['@source'], 'refseq_protein',
external_ids['refseq_protein'])
s.append(
PBB_Core.WDString(external_ids['refseq_protein'],
'P637',
references=[refseq_ref]))
# uniprot id
uniprot_ref = make_ref_source(record['uniprot']['@source'], 'uniprot',
external_ids['uniprot'])
s.append(
PBB_Core.WDString(external_ids['uniprot'],
'P352',
references=[uniprot_ref]))
############
# GO terms
# TODO: https://www.wikidata.org/wiki/Q3460832
############
preprocess_go(record)
print(record)
go_source = record['go']['@source']
go_id_prop = source_ref_id[go_source['_id']]
reference = make_ref_source(go_source, go_id_prop,
external_ids[go_id_prop])
for go_level, go_records in record['go']['@value'].items():
level_wdid = go_props[go_level]
for go_record in go_records:
go_wdid = go_wdid_mapping[go_record['id']]
evidence_wdid = go_evidence_codes[go_record['evidence']]
evidence_statement = PBB_Core.WDItemID(value=evidence_wdid,
prop_nr='P459',
is_qualifier=True)
this_reference = copy.deepcopy(reference)
if add_pubmed:
for pubmed in go_record['pubmed']:
pmid_wdid = PBB_Helpers.PubmedStub(pubmed).create(login)
this_reference.append(
PBB_Core.WDItemID(pmid_wdid, 'P248',
is_reference=True))
s.append(
PBB_Core.WDItemID(go_wdid,
level_wdid,
references=[this_reference],
qualifiers=[evidence_statement]))
############
# statements with no referencable sources (make by hand, for now...)
############
# subclass of protein
s.append(PBB_Core.WDItemID('Q8054', 'P279', references=[ensembl_ref]))
# found in taxon
s.append(
PBB_Core.WDItemID(strain_info['organism_wdid'],
'P703',
references=[ensembl_ref]))
# encodes gene
s.append(PBB_Core.WDItemID(gene_qid, 'P702', references=[ensembl_ref]))
try:
wd_item_protein = PBB_Core.WDItemEngine(
item_name=item_name,
domain='proteins',
data=s,
append_value=['P279'],
fast_run=True,
fast_run_base_filter={
'P352': '',
'P703': strain_info['organism_wdid']
})
wd_item_protein.set_label(item_name)
wd_item_protein.set_description(item_description, lang='en')
wd_item_protein.set_aliases(
[record['symbol']['@value'], record['locus_tag']['@value']])
except Exception as e:
print(e)
PBB_Core.WDItemEngine.log(
"ERROR",
format_msg(record['entrezgene']['@value'], str(e), None,
ENTREZ_PROP))
return
try_write(wd_item_protein, record['entrezgene']['@value'], 'P351', login)
def make_chroms(strain_info, retrieved, login):
chrom_wdid = {}
for chrom_num, genome_id in strain_info['chrom_genomeid_map'].items():
item_name = '{} chromosome {}'.format(strain_info['organism_name'],
chrom_num)
item_description = '{} chromosome'.format(strain_info['organism_type'])
print(item_name)
print(genome_id)
reference = make_ref(retrieved, genome_id)
statements = []
statements.append(
PBB_Core.WDItemID(value='Q37748',
prop_nr='P279',
references=[reference
])) # subclass of chromosome
statements.append(
PBB_Core.WDItemID(value=strain_info['organism_wdid'],
prop_nr='P703',
references=[reference])) # found in taxon
statements.append(
PBB_Core.WDString(value=genome_id,
prop_nr='P2249',
references=[reference])) # genome id
wd_item = PBB_Core.WDItemEngine(item_name=item_name,
domain='chromosome',
data=statements,
append_value=['P279'],
fast_run=True,
fast_run_base_filter={
'P703':
strain_info['organism_wdid'],
'P2249': ''
})
if wd_item.require_write:
print("require write")
wd_item.set_label(item_name)
wd_item.set_description(item_description, lang='en')
try:
msg = "CREATE" if wd_item.create_new_item else "UPDATE"
wd_item.write(login=login)
PBB_Core.WDItemEngine.log(
"INFO",
format_msg(genome_id,
msg,
wd_item.wd_item_id,
external_id_prop='P2249'))
except Exception as e:
print(e)
PBB_Core.WDItemEngine.log(
"ERROR",
format_msg(genome_id,
str(e),
wd_item.wd_item_id,
external_id_prop='P2249'))
else:
chrom_wdid[chrom_num] = wd_item.wd_item_id
PBB_Core.WDItemEngine.log(
"INFO",
format_msg(genome_id,
"SKIP",
wd_item.wd_item_id,
external_id_prop='P2249'))
return chrom_wdid