def build_id_query(self, vid, scopes=None):
_default_scopes = '_id'
scopes = scopes or _default_scopes
if is_str(scopes):
_query = {
"match": {
scopes: {
"query": "{}".format(vid),
"operator": "and"
}
}
}
elif is_seq(scopes):
_query = {
"multi_match": {
"query": "{}".format(vid),
"fields": scopes,
"operator": "and"
}
}
else:
raise ValueError('"scopes" cannot be "%s" type'.format(
type(scopes)))
_q = {"query": _query}
self._query_options.pop(
"query",
None) # avoid "query" be overwritten by self.query_options
_q.update(self._query_options)
return _q
def build_id_query(self, vid, scopes=None):
# _default_scopes = [
# '_id',
# 'rsid', "dbnsfp.rsid", "dbsnp.rsid", "evs.rsid", "mutdb.rsid" # for rsid
# "dbsnp.gene.symbol", 'evs.gene.symbol', 'clinvar.gene.symbol',
# 'dbnsfp.genename', "cadd.gene.genename", "docm.genename", # for gene symbols
# ]
_default_scopes = '_id'
scopes = scopes or _default_scopes
if is_str(scopes):
_query = {
"match": {
scopes: {
"query": "{}".format(vid),
"operator": "and"
}
}
}
elif is_seq(scopes):
_query = {
"multi_match": {
"query": "{}".format(vid),
"fields": scopes,
"operator": "and"
}
}
else:
raise ValueError('"scopes" cannot be "%s" type'.format(type(scopes)))
_q = {"query": _query}
self._query_options.pop("query", None) # avoid "query" be overwritten by self.query_options
_q.update(self._query_options)
return _q
def build_id_query(self, vid, scopes=None):
_default_scopes = '_id'
scopes = scopes or _default_scopes
if is_str(scopes):
_query = {
"match": {
scopes: {
"query": "{}".format(vid),
"operator": "and"
}
}
}
elif is_seq(scopes):
_query = {
"multi_match": {
"query": "{}".format(vid),
"fields": scopes,
"operator": "and"
}
}
else:
raise ValueError('"scopes" cannot be "%s" type'.format(type(scopes)))
_q = {"query": _query}
self._query_options.pop("query", None) # avoid "query" be overwritten by self.query_options
_q.update(self._query_options)
return _q
def build_id_query(self, vid, scopes=None):
# _default_scopes = [
# '_id',
# 'rsid', "dbnsfp.rsid", "dbsnp.rsid", "evs.rsid", "mutdb.rsid" # for rsid
# "dbsnp.gene.symbol", 'evs.gene.symbol', 'clinvar.gene.symbol',
# 'dbnsfp.genename', "cadd.gene.genename", "docm.genename", # for gene symbols
# ]
_default_scopes = '_id'
scopes = scopes or _default_scopes
if is_str(scopes):
_query = {
"match": {
scopes: {
"query": "{}".format(vid),
"operator": "and"
}
}
}
elif is_seq(scopes):
_query = {
"multi_match": {
"query": "{}".format(vid),
"fields": scopes,
"operator": "and"
}
}
else:
raise ValueError('"scopes" cannot be "%s" type'.format(
type(scopes)))
_q = {"query": _query}
self._query_options.pop(
"query",
None) # avoid "query" be overwritten by self.query_options
_q.update(self._query_options)
return _q
def to_number(val):
"""convert an input string to int/float."""
if is_str(val):
try:
return int(val)
except ValueError:
try:
return float(val)
except ValueError:
pass
return val
def to_number(val):
"""convert an input string to int/float."""
if is_str(val):
try:
return int(val)
except ValueError:
try:
return float(val)
except ValueError:
pass
return val
def validate_src(self,
collection,
return_false=False,
return_none=False,
return_true=False,
verbose=False,
flag_invalid=False):
'''Validate hgvs ids from a src collection.'''
return_dict = {
False: return_false,
True: return_true,
None: return_none
}
# read in the collection from mongodb
if is_str(collection):
src = get_src_db()
_coll = src[collection]
else:
_coll = collection
cursor = doc_feeder(_coll, step=10000)
out = {}
print_only = not (return_false or return_none or return_true)
if not print_only:
# output dictionary, three keys: 'false','true','none'
for k in return_dict:
if return_dict[k]:
out[k] = []
# initialize the count
cnt_d = {True: 0, False: 0, None: 0} # cnt_d
# validate each item in the cursor
for item in cursor:
_id = item['_id']
valid = self.validate_hgvs(_id, verbose=verbose)
if valid == False and flag_invalid:
collection.update({"_id": _id},
{'$set': {
"unmatched_ref": "True"
}})
cnt_d[valid] += 1
if return_dict[valid]:
out[valid].append(_id)
# print out counts
print("\n# of VALID HGVS IDs:\t{0}".format(cnt_d[True]))
print("# of INVALID HGVS IDs:\t{0}".format(cnt_d[False]))
print("# of HGVS IDs skipped:\t {0}".format(cnt_d[None]))
out['summary'] = cnt_d
return out
def _cleaned_fields(self, fields):
'''return a cleaned fields parameter.
should be either None (return all fields) or a list fields.
'''
if fields:
if is_str(fields):
if fields.lower() == 'all':
fields = None # all fields will be returned.
else:
fields = [x.strip() for x in fields.split(',')]
else:
fields = self._default_fields
return fields
def _cleaned_fields(self, fields):
'''return a cleaned fields parameter.
should be either None (return all fields) or a list fields.
'''
if fields:
if is_str(fields):
if fields.lower() == 'all':
fields = None # all fields will be returned.
else:
fields = [x.strip() for x in fields.split(',')]
else:
fields = self._default_fields
return fields
def _cleaned_scopes(self, scopes):
'''return a cleaned scopes parameter.
should be either a string or a list of scope fields.
'''
if scopes:
if is_str(scopes):
scopes = [x.strip() for x in scopes.split(',')]
if is_seq(scopes):
scopes = [x for x in scopes if x]
if len(scopes) == 1:
scopes = scopes[0]
else:
scopes = None
else:
scopes = None
return scopes
def _cleaned_scopes(self, scopes):
'''return a cleaned scopes parameter.
should be either a string or a list of scope fields.
'''
if scopes:
if is_str(scopes):
scopes = [x.strip() for x in scopes.split(',')]
if is_seq(scopes):
scopes = [x for x in scopes if x]
if len(scopes) == 1:
scopes = scopes[0]
else:
scopes = None
else:
scopes = None
return scopes
def validate_src(self, collection, return_false=False,
return_none=False, return_true=False, verbose=False, flag_invalid=False, generator=False):
'''Validate hgvs ids from a src collection.'''
return_dict = {
False: return_false,
True: return_true,
None: return_none
}
# read in the collection from mongodb
if is_str(collection):
src = get_src_db()
_coll = src[collection]
else:
_coll = collection
cursor = doc_feeder(_coll, step=10000)
out = {}
print_only = not (return_false or return_none or return_true)
if not print_only:
# output dictionary, three keys: 'false','true','none'
for k in return_dict:
if return_dict[k]:
out[k] = []
# initialize the count
cnt_d = {True: 0, False: 0, None: 0} # cnt_d
# validate each item in the cursor
for item in cursor:
_id = item['_id']
valid = self.validate_hgvs(_id, verbose=verbose)
if valid == False and flag_invalid:
collection.update({"_id": _id}, {'$set':{"unmatched_ref": "True"}})
cnt_d[valid] += 1
if return_dict[valid]:
out[valid].append(_id)
# print out counts
print("\n# of VALID HGVS IDs:\t{0}".format(cnt_d[True]))
print("# of INVALID HGVS IDs:\t{0}".format(cnt_d[False]))
print("# of HGVS IDs skipped:\t {0}".format(cnt_d[None]))
out['summary'] = cnt_d
return out
