def build_database(connection):
"""Build the hpo database
Args:
conn (sqlite3.connect): A database connection object
"""
logger.info("Fetching disease to gene file")
disease_to_genes = getreader('utf-8')(gzip.open(disease_to_genes_path),
errors='replace')
logger.info("Fetching phenotypes file")
phenotype_to_genes = getreader('utf-8')(gzip.open(phenotypes_path),
errors='replace')
logger.info("Fetching genes file")
genes = getreader('utf-8')(gzip.open(genes_path), errors='replace')
logger.info("Parsing phenotypes to gene file")
phenotypes = parse_phenotypes(phenotype_to_genes)
logger.info("Parsing disease to gene file")
diseases = parse_diseases(disease_to_genes)
logger.info("Parsing genes file")
genes = parse_genes(genes)
logger.info("Populating hpo table")
populate_hpo(connection=connection, hpo_terms=phenotypes)
logger.debug("Hpo table populated")
logger.info("Populating disease table")
populate_disease(connection=connection, disease_terms=diseases)
logger.debug("Disease table populated")
logger.info("Populating gene table")
populate_genes(connection=connection, genes=genes)
logger.debug("Gene table populated")
def genes(request):
"""Get hpo terms"""
gene_lines = [
"#chrom\tstart\tstop\tensembl_id\tdescription\thgnc_symbol\thi_score\t"\
"constraint_score",
"2\t203044020\t203044694\tENSG00000227890\tproteasome (prosome,"\
" macropain) subunit, alpha type, 2 pseudogene 3\tPSMA2P3\t\t",
"11\t87040449\t87041094\tENSG00000254582\tproteasome (prosome,"\
" macropain) subunit, alpha type, 2 pseudogene 1\tPSMA2P1\t\t",
"5\t149737202\t149779871\tENSG00000070814\tTreacher Collins-"\
"Franceschetti syndrome 1\tTCOF1\t0.202\t4.98430839027322",
"17\t28442539\t28513493\tENSG00000126653\tnuclear speckle splicing"\
" regulatory protein 1\tNSRP1\t\t2.9732792450519",
]
return parse_genes(gene_lines)
def build_database(connection):
"""Build the hpo database
Args:
conn (sqlite3.connect): A database connection object
"""
logger.info("Fetching disease to gene file")
disease_to_genes = getreader('utf-8')(
gzip.open(disease_to_genes_path), errors='replace')
logger.info("Fetching phenotypes file")
phenotype_to_genes = getreader('utf-8')(
gzip.open(phenotypes_path), errors='replace')
logger.info("Fetching genes file")
genes = getreader('utf-8')(
gzip.open(genes_path), errors='replace')
logger.info("Parsing phenotypes to gene file")
phenotypes = parse_phenotypes(phenotype_to_genes)
logger.info("Parsing disease to gene file")
diseases = parse_diseases(disease_to_genes)
logger.info("Parsing genes file")
genes = parse_genes(genes)
logger.info("Populating hpo table")
populate_hpo(
connection=connection,
hpo_terms=phenotypes
)
logger.debug("Hpo table populated")
logger.info("Populating disease table")
populate_disease(
connection=connection,
disease_terms=diseases
)
logger.debug("Disease table populated")
logger.info("Populating gene table")
populate_genes(
connection=connection,
genes=genes
)
logger.debug("Gene table populated")
