def filter_queryset(self, request, queryset, view):
''' Override this method to request feature locations. '''
try:
filterable = getattr(view, 'filter_fields', [])
filters = dict([(k, v) for k, v in request.GET.items() if k in filterable])
query_str = filters.get('feature', 'PTPN22')
build = self._get_build(filters.get('build', settings.DEFAULT_BUILD))
if query_str is None or query_str == '':
return [ElasticObject(initial={'error': 'No feature name provided.'})]
search_fields = ['id',
'symbol', 'dbxrefs.ensembl',
'region_name']
sources = ['start', 'stop', 'seqid', 'chromosome',
'disease_loci']
idxs = ElasticSettings.getattr('IDX')
MARKER_IDX = ''
if build == ElasticSettings.get_label('MARKER', label='build'):
MARKER_IDX = 'MARKER'
if MARKER_IDX == '':
for idx in idxs:
if 'MARKER' in idx:
if build == ElasticSettings.get_label(idx, label='build'):
MARKER_IDX = idx
(idx, idx_type) = ElasticSettings.idx_names(MARKER_IDX, 'MARKER')
(idx_r, idx_type_r) = ElasticSettings.idx_names('REGION', 'REGION')
(idx_g, idx_type_g) = ElasticSettings.idx_names('GENE', 'GENE')
idx += ',' + idx_r + ',' + idx_g
idx_type += ',' + idx_type_r + ',' + idx_type_g
equery = BoolQuery(must_arr=Query.query_string(query_str, fields=search_fields))
elastic = Search(search_query=ElasticQuery(equery, sources), size=10, idx=idx, idx_type=idx_type)
docs = elastic.search().docs
locs = []
for doc in docs:
if isinstance(doc, RegionDocument):
doc = Region.pad_region_doc(doc)
loc = doc.get_position(build=build).split(':')
pos = loc[1].replace(',', '').split('-')
locs.append(ElasticObject(
{'feature': query_str,
'chr': loc[0],
'start': int(pos[0]),
'end': int(pos[1]) if len(pos) > 1 else int(pos[0]),
'locusString': query_str+" ("+str(loc[1])+")"}))
return locs
except (TypeError, ValueError, IndexError, ConnectionError):
raise Http404
def do_identifier_search(cls, identifiers, user=None):
source_filter = [
'symbol', 'synonyms', "dbxrefs.*", # gene
'id', 'rscurrent', 'rshigh', # marker
'study_id', 'study_name', # study
'region_name', 'marker', "region_id"] # regions
highlight = Highlight(["symbol", "dbxrefs.*", "region", "region_name", "region_id",
"study_id", "study_name", "id", "rscurrent", "rshigh", "marker"])
search_query = ElasticQuery(Query.query_string(" ".join(identifiers), fields=source_filter),
highlight=highlight, sources=source_filter)
search_idx_keys = ['REGION', 'GENE', 'STUDY', 'MARKER']
search_idx_type_keys = ['REGION', 'GENE', 'STUDY', 'MARKER']
idx_all = [ElasticSettings.idx_names(idx, idx_type=idx_type) for idx, idx_type in zip(search_idx_keys,
search_idx_type_keys)]
idx_dict = dict(idx_all)
search_idx = ','.join(idx_dict.keys())
search_idx_types = ','.join(idx_dict.values())
elastic = Search(search_query=search_query, idx=search_idx, idx_type=search_idx_types)
gene_dict = {}
region_dict = {}
marker_dict = {}
study_dict = {}
docs = elastic.search().docs
for doc in docs:
existing_feature_list = []
idx = getattr(doc, '_meta')['_index']
idx_type = getattr(doc, '_meta')['_type']
doc_id = doc.doc_id()
highlight = doc.highlight()
if highlight is not None:
pattern = ".*?<em>(.*?)</em>.*"
result = re.match(pattern, str(highlight))
if result is not None:
highlight_hit = result.group(1)
if idx_type == "studies":
feature_id = getattr(doc, "study_id")
if highlight_hit not in study_dict:
study_dict[highlight_hit] = {}
if feature_id in study_dict[highlight_hit]:
existing_feature_list = study_dict[highlight_hit]
existing_feature_list.append(feature_id)
study_dict[highlight_hit] = existing_feature_list
if idx_type == "gene":
feature_id = doc_id
if highlight_hit not in gene_dict:
gene_dict[highlight_hit] = {}
if feature_id in gene_dict[highlight_hit]:
existing_feature_list = gene_dict[highlight_hit]
existing_feature_list.append(feature_id)
gene_dict[highlight_hit] = existing_feature_list
if idx_type == "marker":
feature_id = getattr(doc, "id")
if highlight_hit not in marker_dict:
marker_dict[highlight_hit] = {}
if feature_id in marker_dict[highlight_hit]:
existing_feature_list = marker_dict[highlight_hit]
existing_feature_list.append(feature_id)
marker_dict[highlight_hit] = existing_feature_list
if idx_type == "region":
feature_id = getattr(doc, "region_id")
if highlight_hit not in region_dict:
region_dict[highlight_hit] = {}
if feature_id in region_dict[highlight_hit]:
existing_feature_list = region_dict[highlight_hit]
existing_feature_list.append(feature_id)
region_dict[highlight_hit] = existing_feature_list
all_result_dict = {}
all_result_dict['gene'] = gene_dict
all_result_dict['marker'] = marker_dict
all_result_dict['region'] = region_dict
all_result_dict['study'] = study_dict
original_list = [_id.lower() for _id in identifiers]
result_list = list(study_dict.keys()) + list(gene_dict.keys()) + list(marker_dict.keys()) + \
list(region_dict.keys())
result_list = [_id.lower() for _id in result_list]
diff_list = set(original_list) - set(result_list)
all_result_dict['missing'] = list(diff_list)
return all_result_dict
