def get_protein_domain_data(uniprot_id):
query_tpl = 'SELECT interpro_id, name, database, start, end ' \
'FROM {interpro_table} ' \
'WHERE uniprot_accession=?'
query = query_tpl.format(interpro_table=INTERPRO_TABLE)
log.debug("INTERPRO SQL: " + query)
values = [uniprot_id]
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
res = {k: row[k] for k in row.keys()}
items.append(res)
cursor.close()
db.close()
if len(items) == 0:
raise InterProEntryNotFound(uniprot_id)
return items
def get_mutation_data(tumor_type_array, gene):
# Generate the 'IN' statement string: (%s, %s, ..., %s)
tumor_stmt = ', '.join(['?' for tumor in tumor_type_array])
query_tpl = 'SELECT Cancer AS tumor_type, gene, protein_ID, tumor_sample AS patient_id, mutation_type, aa_change, aa_location, aa1, aa2 ' \
'FROM {mutation_table} ' \
'WHERE gene=? AND Cancer IN ({tumor_stmt})'
query = query_tpl.format(mutation_table=MUTATION_TABLE,
tumor_stmt=tumor_stmt)
log.debug("MUTATION SQL: " + query)
values = [gene]
values.extend(tumor_type_array)
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
cluster = parse_cluster(row)
items.append(cluster)
cursor.close()
db.close()
log.debug("Found mutation rows: {num}".format(num=len(items)))
return items
def get_protein_domain_data(uniprot_id):
query_tpl = 'SELECT interpro_id, name, database, start, end ' \
'FROM {interpro_table} ' \
'WHERE uniprot_accession=?'
query = query_tpl.format(interpro_table=INTERPRO_TABLE)
log.debug("INTERPRO SQL: " + query)
values = [uniprot_id]
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
res = {k: row[k] for k in row.keys()}
items.append(res)
cursor.close()
db.close()
if len(items) == 0:
raise InterProEntryNotFound(uniprot_id)
return items
def get_uniprot_data(uniprot_id):
"""
:param uniprot_id:
:return:
"""
query_tpl = 'SELECT primary_accession, entry_name, length, protein_name ' \
'FROM {uniprot_table} ' \
'WHERE primary_accession=?'
query = query_tpl.format(uniprot_table=UNIPROT_TABLE)
_log.debug("UNIPROT SQL: " + query)
values = [uniprot_id]
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
res = {k: row[k] for k in row.keys()}
items.append(res)
cursor.close()
db.close()
if len(items) == 0:
return None
return items[0]
def get_cluster_data(tumor_type_array, gene):
# Generate the 'IN' statement string: (%s, %s, ..., %s)
tumor_stmt = ', '.join(['?' for tumor in tumor_type_array])
query_tpl = 'SELECT cancer as tumor_type, ct.gene, ct.cluster, missense_mutations, nonsense_mutations, silent_mutations, cs.cluster_score '\
'FROM {cluster_table} ct ' \
'LEFT OUTER JOIN {cluster_score_table} cs ON ct.gene = cs.gene AND ct.cluster = cs.cluster ' \
'WHERE ct.gene=? AND ct.cancer IN ({tumor_stmt}) '
query = query_tpl.format(cluster_table=CLUSTER_TABLE,
cluster_score_table=CLUSTER_SCORE_TABLE,
tumor_stmt=tumor_stmt)
log.debug("CLUSTER SQL: " + query)
values = [gene]
values.extend(tumor_type_array)
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
cluster = parse_cluster(row)
items.append(cluster)
cursor.close()
db.close()
return items
def get_mutation_data(tumor_type_array, gene):
# Generate the 'IN' statement string: (%s, %s, ..., %s)
tumor_stmt = ', '.join(['?' for tumor in tumor_type_array])
query_tpl = 'SELECT Cancer AS tumor_type, gene, protein_ID, tumor_sample AS patient_id, mutation_type, aa_change, aa_location, aa1, aa2 ' \
'FROM {mutation_table} ' \
'WHERE gene=? AND Cancer IN ({tumor_stmt})'
query = query_tpl.format(mutation_table=MUTATION_TABLE, tumor_stmt=tumor_stmt)
log.debug("MUTATION SQL: " + query)
values = [gene]
values.extend(tumor_type_array)
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
cluster = parse_cluster(row)
items.append(cluster)
cursor.close()
db.close()
log.debug("Found mutation rows: {num}".format(num=len(items)))
return items
def get_cluster_data(tumor_type_array, gene):
# Generate the 'IN' statement string: (%s, %s, ..., %s)
tumor_stmt = ', '.join(['?' for tumor in tumor_type_array])
query_tpl = 'SELECT cancer as tumor_type, ct.gene, ct.cluster, missense_mutations, nonsense_mutations, silent_mutations, cs.cluster_score '\
'FROM {cluster_table} ct ' \
'LEFT OUTER JOIN {cluster_score_table} cs ON ct.gene = cs.gene AND ct.cluster = cs.cluster ' \
'WHERE ct.gene=? AND ct.cancer IN ({tumor_stmt}) '
query = query_tpl.format(cluster_table=CLUSTER_TABLE, cluster_score_table=CLUSTER_SCORE_TABLE, tumor_stmt=tumor_stmt)
log.debug("CLUSTER SQL: " + query)
values = [gene]
values.extend(tumor_type_array)
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
cluster = parse_cluster(row)
items.append(cluster)
cursor.close()
db.close()
return items
def get_mutation_data_summary_for_gene(gene):
query_tpl = 'SELECT Cancer as tumor_type, gene, protein_ID, tumor_sample AS patient_id, mutation_type, aa_change, aa_location, aa1, aa2 ' \
'FROM {mutation_table} ' \
'WHERE gene=?'
query = query_tpl.format(mutation_table=MUTATION_TABLE)
log.debug("MUTATION SUMMARY SQL: " + query)
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, (gene, ))
items = []
for row in cursor.fetchall():
cluster = parse_cluster(row)
items.append(cluster)
cursor.close()
db.close()
log.debug("Found mutation summary rows: {num}".format(num=len(items)))
return items
def get_mutation_data_summary_for_gene(gene):
query_tpl = 'SELECT Cancer as tumor_type, gene, protein_ID, tumor_sample AS patient_id, mutation_type, aa_change, aa_location, aa1, aa2 ' \
'FROM {mutation_table} ' \
'WHERE gene=?'
query = query_tpl.format(mutation_table=MUTATION_TABLE)
log.debug("MUTATION SUMMARY SQL: " + query)
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, (gene, ))
items = []
for row in cursor.fetchall():
cluster = parse_cluster(row)
items.append(cluster)
cursor.close()
db.close()
log.debug("Found mutation summary rows: {num}".format(num=len(items)))
return items
def get_pathway_data(tumor_type, gene, cluster):
query_tpl = 'SELECT cancer as tumor_type, gene, cluster, pathway_name, pval, fdr, wl.url ' \
'FROM {pathway_assoc_table} a ' \
'LEFT OUTER JOIN pathway_web_links wl ON a.pathway_name=wl.pathway ' \
'WHERE a.gene=? AND a.cancer=? AND a.cluster=? '
query = query_tpl.format(pathway_assoc_table=PATHWAY_ASSOC_TABLE)
log.debug("PATHWAY_ASSOC SQL: " + query)
values = [gene, tumor_type, cluster]
db = sql_connection()
cursor = db.cursor()
cursor.execute(query, tuple(values))
items = []
for row in cursor.fetchall():
items.append(row)
log.debug("Found {0} pathway associations.".format(len(items)))
cursor.close()
db.close()
return items