def default_string_query(self, q, options):
# genomic interval query
pattern = r'chr(?P<chrom>\w+):(?P<gstart>[0-9,]+)-(?P<gend>[0-9,]+)'
match = re.search(pattern, q)
if match: # (chr, gstart, gend)
d = match.groupdict()
if q.startswith('hg19.'):
# support hg19 for human (default is hg38)
d['assembly'] = 'hg19'
if q.startswith('mm9.'):
# support mm9 for mouse (default is mm10)
d['assembly'] = 'mm9'
search = AsyncSearch().from_dict(interval(**d))
# query_string query
elif q.startswith('"') and q.endswith('"') or \
any(map(q.__contains__, (':', '~', ' AND ', ' OR ', 'NOT '))):
search = AsyncSearch().query("query_string",
query=q,
default_operator="AND",
auto_generate_phrase_queries=True)
# wildcard query
elif '*' in q or '?' in q:
search = AsyncSearch().from_dict(wildcard(q))
else: # default query
search = AsyncSearch().from_dict(dismax(q))
search = self._extra_query_options(search, options)
return search
def _build_string_query(self, q, options):
""" q + options -> query object
options:
userquery
"""
assert isinstance(q, str)
search = AsyncSearch()
userquery = options.userquery or ''
if not q: # same empty q behavior as that of ES.
search = search.query("match_none")
elif self.user_query.has_query(userquery):
userquery_ = self.user_query.get_query(userquery, q=q)
search = search.query(userquery_)
else: # customization here
search = self.default_string_query(q, options)
if self.user_query.has_filter(userquery):
userfilter = self.user_query.get_filter(userquery)
search = search.filter(userfilter)
return search
def default_string_query(self, q, options):
search = AsyncSearch()
if q == '__all__':
search = search.query()
elif q == '__any__' and self.allow_random_query:
search = search.query('function_score', random_score={})
else: # elasticsearch default
query = {
"query": {
"function_score": {
"query": {"dis_max": {"queries": [
{"term": {"_id": {"value": q, "boost": 15.0}}},
{"term": {"label.raw": {"value": q, "boost": 10.0}}},
{"term": {"_meta.username": {"value": q}}}, # for dataset
{"term": {"name": {"value": q}}},
{"match": {"parent_classes": {"query": q}}},
{"prefix": {"label": {"value": q}}},
{"query_string": {"query": q}}
]}},
"functions": [
{"filter": {"term": {"namespace": "schema"}}, "weight": 0.5},
{"filter": {"term": {"prefix.raw": "schema"}}, "weight": 0.5},
{"filter": {"match": {"parent_classes": "bts:BiologicalEntity"}}, "weight": 1.5}
]
}
}
}
search = AsyncSearch()
search = search.update_from_dict(query)
search = search.params(rest_total_hits_as_int=True)
return search
def default_string_query(self, q, options):
query = {
"query": {
"query_string": {
"query":
q,
"fields":
["name^4", "interventions.name^3", "description", "all"]
}
}
}
search = AsyncSearch()
search = search.update_from_dict(query)
return search
async def execute_pipeline(self, *args, **kwargs):
try:
graph_query = GraphQuery.from_dict(self.args_json)
es_query = self._to_es_query(graph_query)
if graph_query.can_reverse():
graph_query.reverse()
es_query_rev = self._to_es_query(graph_query)
es_query = es_query | es_query_rev
# it's sent in one query so that parameters like size is still meaningful
_query = AsyncSearch().query(es_query)
_res = await self.pipeline.execute(_query, dotdict())
res = self.pipeline.transform(_res, dotdict())
# TODO additional transformation, like double reversal in result.
except GraphObjectError as exc:
raise BadRequest(reason=str(exc))
except Exception as exc:
raise HTTPError(str(exc))
self.finish(res)
def default_match_query(self, q, scopes, options):
'''
Override this to customize default match query.
By default it implements a multi_match query.
'''
if isinstance(q, (str, int, float)):
query = Q('multi_match',
query=str(q),
operator="and",
fields=scopes,
lenient=True)
elif isinstance(q, list):
if not isinstance(scopes, list):
raise TypeError(scopes)
if len(q) != len(scopes):
raise ValueError(q)
query = Q() # combine conditions
for _q, _scopes in zip(q, scopes):
query = query & Q('multi_match',
query=_q,
operator="and",
fields=_scopes,
lenient=True)
else: # invalid
raise TypeError(q)
return AsyncSearch().query(query)
def default_string_query(self, q, options):
"""
Override this to customize default string query.
By default it implements a query string query.
"""
search = AsyncSearch()
if q == '__all__':
search = search.query()
elif q == '__any__' and self.allow_random_query:
search = search.query('function_score', random_score={})
else: # elasticsearch default
search = search.query("query_string", query=str(q))
return search
def _extra_query_options(self, search, options):
search = AsyncSearch().query(
"function_score",
query=search.query,
functions=[
{"filter": {"term": {"taxid": 9606}}, "weight": "1.55"}, # human
{"filter": {"term": {"taxid": 10090}}, "weight": "1.3"}, # mouse
{"filter": {"term": {"taxid": 10116}}, "weight": "1.1"}, # rat
], score_mode="first")
if options.species:
if 'all' in options.species:
pass
elif not all(isinstance(string, str) for string in options.species):
raise BadRequest(reason="species must be strings or integer strings.")
elif not all(string.isnumeric() for string in options.species):
raise BadRequest(reason="cannot map some species to taxids.")
else:
search = search.filter('terms', taxid=options.species)
if options.aggs and options.species_facet_filter:
search = search.post_filter('terms', taxid=options.species_facet_filter)
return search
def build_string_query(self, q, options):
""" q + options -> query object
options:
userquery
"""
assert isinstance(q, str)
search = AsyncSearch()
userquery = options.userquery or ''
if self.user_query.has_query(userquery):
userquery_ = self.user_query.get_query(userquery, q=q)
search = search.query(userquery_)
elif q == '__all__':
search = search.query()
elif q == '__any__' and self.allow_random_query:
search = search.query('function_score', random_score={})
else: # customization here
search = self.default_string_query(q, options)
if self.user_query.has_filter(userquery):
userfilter = self.user_query.get_filter(userquery)
search = search.filter(userfilter)
return search
def default_string_query(self, q, options):
'''
Override this to customize default string query.
By default it implements a query string query.
'''
## for extra query types:
#
# if q == 'case_1':
# return case_1(q)
# elif q == 'case_2':
# return case_2(q)
#
# return default_case(q)
return AsyncSearch().query("query_string", query=str(q))
def build_lineage_query(_id, options):
search = AsyncSearch()
search = search.query('match', lineage=_id)
if options.has_gene:
search = search.query('match', has_gene=options.has_gene)
max_taxid_count = 10000
search = search.params(size=max_taxid_count)
search = search.params(_source='_id')
return search
def default_string_query(self, q, options):
return AsyncSearch().from_dict({
"query": {
"dis_max": {
"queries": [{
"query_string": {
"query": q,
"fields": ["name^6", "description^3"]
}
}, {
"query_string": {
"query": q
}
}]
}
}
})
def default_string_query(self, q, options):
match = self._parse_interval_query(q)
if match: # interval query
search = AsyncSearch()
if match['query']:
search = search.query("query_string", query=match['query'])
search = search.filter('match', chrom=match['chr'])
assembly = 'hg38' if options.assembly == 'hg38' else 'hg19'
search = search.filter(
'range', **{assembly + ".start": {
"lte": match['gend']
}})
search = search.filter(
'range', **{assembly + ".end": {
"gte": match['gstart']
}})
else: # default query
search = AsyncSearch().query("query_string", query=q)
return search
def _extra_query_options(self, search, options):
search = AsyncSearch().query(
"function_score",
query=search.query,
functions=[
{"filter": {"term": {"name": "pseudogene"}}, "weight": "0.5"}, # downgrade
{"filter": {"term": {"taxid": 9606}}, "weight": "1.55"},
{"filter": {"term": {"taxid": 10090}}, "weight": "1.3"},
{"filter": {"term": {"taxid": 10116}}, "weight": "1.1"},
], score_mode="first")
if options.entrezonly:
search = search.filter('exists', field="entrezgene")
if options.ensemblonly:
search = search.filter('exists', field="ensembl.gene")
if options.missing:
for field in options.missing:
search = search.exclude('exists', field=field)
if options.exists:
for field in options.exists:
search = search.filter('exists', field=field)
if options.species:
if 'all' in options.species:
pass # do not apply any filters
elif not all(isinstance(string, str) for string in options.species):
raise BadRequest(reason="species must be strings or integer strings.")
elif not all(string.isnumeric() for string in options.species):
raise BadRequest(reason="cannot map some species to taxids.")
else: # filter by taxid numeric strings
search = search.filter('terms', taxid=options.species)
if options.aggs and options.species_facet_filter:
search = search.post_filter('terms', taxid=options.species_facet_filter)
return search
def default_string_query(self, q, options):
search = AsyncSearch()
q = q.strip()
if q == '__all__':
search = search.query()
elif q == '__any__' and self.allow_random_query:
search = search.query('function_score', random_score={})
# elasticsearch query string syntax
elif ":" in q or " AND " in q or " OR " in q:
search = search.query('query_string', query=q)
# term search
elif q.startswith('"') \
and q.endswith('"'):
query = {
"query": {
"dis_max": {
"queries": [
{
"term": {
"_id": {
"value": q.strip('"'),
"boost": 5
}
}
},
{
"term": {
"_meta.slug": {
"value": q.strip('"'),
"boost": 5
}
}
},
{
"match": {
"info.title": {
"query": q,
"boost": 1.5,
"operator": "AND"
}
}
},
{
"query_string": {
"query": q,
"default_operator": "AND"
}
} # base score
]
}
}
}
search = AsyncSearch()
search = search.update_from_dict(query)
else: # simple text search
query = {
"query": {
"dis_max": {
"queries": [
{
"term": {
"_id": {
"value": q,
"boost": 5
}
}
},
{
"term": {
"_meta.slug": {
"value": q,
"boost": 5
}
}
},
{
"match": {
"info.title": {
"query": q,
"boost": 1.5
}
}
},
{
"term": {
"servers.url": {
"value": q,
"boost": 1.1
}
}
},
# ---------------------------------------------
{
"query_string": {
"query": q
}
}, # base score
# ---------------------------------------------
{
"wildcard": {
"info.title": {
"value": q + "*",
"boost": 0.8
}
}
},
{
"wildcard": {
"info.description": {
"value": q + "*",
"boost": 0.5
}
}
},
]
}
}
}
search = AsyncSearch()
search = search.update_from_dict(query)
search = search.params(rest_total_hits_as_int=True)
search = search.source(exclude=['_raw'], include=options._source)
if options.authors: # '"Chunlei Wu"'
search = search.filter('terms',
info__contact__name__raw=options.authors)
if options.tags: # '"chemical", "drug"'
search = search.filter('terms', tags__name__raw=options.tags)
return search
def _extra_query_options(self, search, options):
search = AsyncSearch().query(
"function_score",
query=search.query,
functions=[
{
"filter": {
"term": {
"name": "pseudogene"
}
},
"weight": "0.5"
}, # downgrade
{
"filter": {
"term": {
"taxid": 9606
}
},
"weight": "1.55"
},
{
"filter": {
"term": {
"taxid": 10090
}
},
"weight": "1.3"
},
{
"filter": {
"term": {
"taxid": 10116
}
},
"weight": "1.1"
},
],
score_mode="first")
if options.entrezonly:
search = search.filter('exists', field="entrezgene")
if options.ensemblonly:
search = search.filter('exists', field="ensembl.gene")
if options.missing:
for field in options.missing:
search = search.exclude('exists', field=field)
if options.exists:
for field in options.exists:
search = search.filter('exists', field=field)
if options.species:
if 'all' not in options.species: # TODO
search = search.filter('terms', taxid=options.species)
if options.aggs and options.species_facet_filter:
search = search.post_filter('terms',
taxid=options.species_facet_filter)
return search