def _collapse_region_docs(cls, docs):
''' If the document is a hit then find parent region; pad all regions for build_info.'''
hits = [doc for doc in docs if doc.type() == 'hits']
regions = [doc for doc in docs if doc.type() == 'region']
if len(hits) > 0:
regions = Region.hits_to_regions(hits)
return regions
def _collapse_region_docs(docs):
''' If the document is a hit then find parent region; pad all regions for build_info.'''
hits = [doc for doc in docs if doc.type() == 'hits']
regions = [doc for doc in docs if doc.type() == 'region']
if len(hits) > 0:
regions = Region.hits_to_regions(hits)
for doc in hits:
docs.remove(doc)
regions = [Region.pad_region_doc(doc) for doc in regions]
for doc in regions:
if doc in docs:
docs.remove(doc)
docs.append(doc)
return len(docs)
def get_overlapping_features(self, build, seqid, start, end):
features = []
study_hits = self.get_overlapping_hits(build, seqid, start, end)
if len(study_hits) == 0:
return features
regions = Region.hits_to_regions(study_hits)
for doc in regions:
doc = Region.pad_region_doc(doc)
loc = doc.get_position(build=build).split(':')
pos = loc[1].replace(',', '').split('-')
feature = {
'name': doc.get_name(),
'id': doc.doc_id(),
'chr': loc[0],
'start': int(pos[0]),
'end': int(pos[1]) if len(pos) > 1 else int(pos[0]),
'strand': doc.get_strand_as_int(),
'attributes': {}
}
loci = doc.get_disease_loci()
sub_features = []
for locus in loci:
attributes = {}
loc = locus.get_position(build=build).split(':')
pos = loc[1].replace(',', '').split('-')
sub_feature = {
'name': locus.get_name(),
'id': locus.doc_id(),
'chr': loc[0],
'start': int(pos[0]),
'end': int(pos[1]) if len(pos) > 1 else int(pos[0]),
'strand': locus.get_strand_as_int()
}
if hasattr(locus, "disease") and getattr(locus, "disease") is not None:
attributes['disease'] = getattr(locus, "disease")
sub_feature['attributes'] = attributes
sub_features.append(sub_feature)
feature['sub_features'] = sub_features
features.append(ElasticObject(feature))
return features
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 _collapse_region_docs(docs):
''' If the document is a hit then find parent region; pad all regions for build_info.'''
hits = [doc for doc in docs if doc.type() == 'hits']
regions = [doc for doc in docs if doc.type() == 'region']
if len(hits) > 0:
regions = Region.hits_to_regions(hits)
for doc in hits:
disease_locus = getattr(doc, "disease_locus")
for region in regions:
disease_loci = getattr(region, "disease_loci")
if disease_locus in disease_loci:
region.__dict__['_meta']['highlight'] = doc.highlight()
docs.remove(doc)
regions = [Region.pad_region_doc(doc) for doc in regions]
for doc in regions:
if doc in docs:
docs.remove(doc)
docs.append(doc)
return len(docs)
def get_context_data(self, **kwargs):
context = super(RegionView, self).get_context_data(**kwargs)
region = kwargs['region'] if 'region' in kwargs else self.request.GET.get('r')
if region is None:
messages.error(self.request, 'No region given.')
raise Http404()
if re.search("[p|q]\d+", region.lower()) is None:
regionDocs = Region.loci_to_regions([region.lower()])
if len(regionDocs) == 0:
messages.error(self.request, 'No region found.')
raise Http404()
region = regionDocs[0].doc_id()
return RegionView.get_region(self.request, region, context)
def region_page(request, region):
''' Renders a region page. '''
if region is None:
messages.error(request, 'No region given.')
raise Http404()
query = ElasticQuery(Query.ids(region.split(',')))
elastic = Search(query, idx=ElasticSettings.idx('REGION', 'REGION'), size=5)
res = elastic.search()
if res.hits_total == 0:
messages.error(request, 'Region(s) '+region+' not found.')
elif res.hits_total < 9:
names = ', '.join([getattr(doc, 'region_name') for doc in res.docs])
REGIONS = [Region.pad_region_doc(doc) for doc in res.docs]
context = {'features': REGIONS, 'title': names}
return render(request, 'region/index.html', context, content_type='text/html')
raise Http404()
def get_region(cls, request, region, context):
if region is None:
messages.error(request, 'No region given.')
raise Http404()
query = ElasticQuery(Query.ids(region.split(',')))
elastic = Search(query, idx=ElasticSettings.idx('REGION', 'REGION'), size=5)
res = elastic.search()
if res.hits_total == 0:
messages.error(request, 'Region(s) '+region+' not found.')
elif res.hits_total < 9:
context['features'] = [Region.pad_region_doc(doc) for doc in res.docs]
fids = [doc.doc_id() for doc in res.docs]
criteria_disease_tags = RegionView.criteria_disease_tags(request, fids)
context['criteria'] = criteria_disease_tags
context['title'] = ', '.join([getattr(doc, 'region_name') for doc in res.docs])
return context
raise Http404()
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])
ftype = filters.get('ftype').upper()
build = self._get_build(filters.get('build', settings.DEFAULT_BUILD))
if ftype is None or ftype == '':
return [ElasticObject(initial={'error': 'No feature type provided.'})]
idx = 'REGION' if ftype == 'ASSOC_SNP' else ftype
idx_type = 'STUDY_HITS' if idx == 'REGION' else ftype
print(idx_type)
doc_class_str = ElasticSettings.get_label(idx, idx_type, label='class')
doc_class = import_string(doc_class_str) if doc_class_str is not None else None
if ftype == 'REGION':
features = doc_class.get_overlapping_features(build, filters.get('chr').replace('chr', ''),
filters.get('start'), filters.get('end'))
return features
if ftype == 'ASSOC_SNP':
hits = doc_class.get_overlapping_hits(build, filters.get('chr').replace('chr', ''),
filters.get('start'), filters.get('end'))
features = Region.get_immune_snps(hits)
return features
sources = ['start', 'stop', 'seqid', 'chromosome', 'strand',
'biotype', 'giestain', 'name', 'symbol', 'id', 'ref', 'alt']
seqid_param = 'seqid'
end_param = 'stop'
start_param = 'start'
if ftype == 'GENE':
seqid_param = 'chromosome'
elastic = Search.range_overlap_query(filters.get('chr').replace('chr', ''), filters.get('start'),
filters.get('end'), idx=ElasticSettings.idx(idx, idx_type), size=10000,
seqid_param=seqid_param, end_param=end_param, start_param=start_param,
field_list=sources)
docs = elastic.search().docs
features = []
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('-')
attributes = {}
feature = {
'name': doc.get_name(),
'id': doc.doc_id(),
'chr': loc[0],
'start': int(pos[0]),
'end': int(pos[1]) if len(pos) > 1 else int(pos[0]),
'strand': doc.get_strand_as_int()
}
if hasattr(doc, "biotype") and getattr(doc, "biotype") is not None:
attributes["biotype"] = getattr(doc, "biotype")
if hasattr(doc, "giestain") and getattr(doc, "giestain") is not None:
attributes["stain"] = getattr(doc, "giestain")
if hasattr(doc, "ref") and hasattr(doc, "alt") and getattr(doc, "ref") is not None and getattr(doc, "alt") is not None:
attributes["alleles"] = getattr(doc, "ref")+"/"+getattr(doc, "alt")
feature['attributes'] = attributes
features.append(ElasticObject(feature))
return features
except (ConnectionError) as err:
logger.error(err)
raise Http404
