def gene_mapping(cls, idx, idx_type, test_mode=False):
''' Load the mapping for the gene index. '''
props = MappingProperties(idx_type)
props.add_property("symbol", "string", analyzer="full_name") \
.add_property("synonyms", "string", analyzer="full_name") \
.add_property("chromosome", "string") \
.add_property("source", "string") \
.add_property("start", "long") \
.add_property("stop", "long") \
.add_property("strand", "string") \
.add_property("description", "string") \
.add_property("biotype", "string") \
.add_property("pmids", "string") \
.add_property("suggest", "completion",
index_analyzer="full_name", search_analyzer="full_name")
dbxref_props = cls._get_nested_prop("dbxrefs", "ensembl")
ortholog_props = MappingProperties("orthologs")
ortholog_props.add_properties(cls._get_nested_prop("mmusculus", "ensembl"))
ortholog_props.add_properties(cls._get_nested_prop("rnorvegicus", "ensembl"))
dbxref_props.add_properties(ortholog_props)
props.add_properties(dbxref_props)
''' create index and add mapping '''
load = Loader()
options = {"indexName": idx, "shards": 5}
if not test_mode:
load.mapping(props, idx_type, analyzer=Loader.KEYWORD_ANALYZER, **options)
return props
def _load_interaction_mappings(cls, section):
'''Load the mappings for interactions index type'''
interaction_mapping = MappingProperties("interactions", "gene")
interaction_mapping.add_property("interactors", "object")
interaction_mapping.add_property("interaction_source", "string")
load = Loader()
idx = section['index']
options = {"indexName": idx, "shards": 1}
status = load.mapping(interaction_mapping, "interactions", analyzer=Loader.KEYWORD_ANALYZER, **options)
return status
def _create_rs_merge_mapping(self, idx_type, **options):
''' Create the mapping for rs index '''
props = MappingProperties(idx_type)
props.add_property("rshigh", "string", index="not_analyzed") \
.add_property("rslow", "string", index="not_analyzed") \
.add_property("build_id", "integer", index="no") \
.add_property("orien", "integer", index="no") \
.add_property("create_time", "date", index="no", property_format="yyyy-MM-dd HH:mm:ss.SSS") \
.add_property("last_updated_time", "date", index="no", property_format="yyyy-MM-dd HH:mm:ss.SSS") \
.add_property("rscurrent", "string", index="not_analyzed") \
.add_property("orien2current", "string", index="no") \
.add_property("notes", "string", index="no")
self.mapping(props, idx_type, **options)
return props
def _create_gff_mapping(self, idx_type, **options):
''' Create the mapping for gff index '''
props = MappingProperties(idx_type)
props.add_property("seqid", "string", index="not_analyzed") \
.add_property("source", "string") \
.add_property("type", "string", index="not_analyzed") \
.add_property("start", "integer", index="not_analyzed") \
.add_property("end", "integer", index="not_analyzed") \
.add_property("score", "string", index="no") \
.add_property("strand", "string", index="no") \
.add_property("phase", "string", index="no") \
.add_property("attr", "object")
self.mapping(props, idx_type, **options)
return props
def mapping(cls, idx, idx_type):
''' Load the mapping for the chromosome type in the bands index.
seqid - chromosome
length - sequence length
'''
props = MappingProperties(idx_type)
props.add_property("seqid", "string", index="not_analyzed") \
.add_property("length", "integer")
''' create index and add mapping '''
load = Loader()
options = {"indexName": idx, "shards": 1}
load.mapping(props, idx_type, analyzer=Loader.KEYWORD_ANALYZER, **options)
return props
def gene_history_mapping(cls, idx, idx_type, test_mode=False):
''' Load the mapping for the gene index. '''
props = MappingProperties(idx_type)
props.add_property("geneid", "integer") \
.add_property("discontinued_geneid", "integer") \
.add_property("discontinued_symbol", "string", analyzer="full_name") \
.add_property("discontinue_date", "date")
''' create index and add mapping '''
load = Loader()
options = {"indexName": idx, "shards": 5}
if not test_mode:
load.mapping(props, idx_type, analyzer=Loader.KEYWORD_ANALYZER, **options)
return props
def _create_mapping(self, **options):
''' Create the mapping for gene names indexing '''
props = MappingProperties("gene")
props.add_property("gene_symbol", "string", analyzer="full_name") \
.add_property("biotype", "string") \
.add_property("synonyms", "string", analyzer="full_name") \
.add_property("hgnc", "string") \
.add_property("dbxrefs", "object") \
.add_property("organism", "string")
featureloc_props = MappingProperties("featureloc")
featureloc_props.add_property("start", "integer") \
.add_property("end", "integer") \
.add_property("seqid", "string") \
.add_property("build", "string")
props.add_properties(featureloc_props)
''' create index and add mapping '''
self.mapping(props, 'gene', analyzer=self.KEYWORD_ANALYZER, **options)
def _create_snp_mapping(self, idx_type, **options):
''' Create the mapping for snp index '''
props = MappingProperties(idx_type)
props.add_property("seqid", "string", index="not_analyzed") \
.add_property("start", "integer", index="not_analyzed") \
.add_property("id", "string", analyzer="full_name") \
.add_property("ref", "string", index="no") \
.add_property("alt", "string", index="no") \
.add_property("qual", "string", index="no") \
.add_property("filter", "string", index="no") \
.add_property("info", "string", index="no") \
.add_property("suggest", "completion", analyzer="full_name")
tags = MappingProperties("tags")
tags.add_property("weight", "integer", index="not_analyzed")
props.add_properties(tags)
self.mapping(props,
idx_type,
analyzer=Loader.KEYWORD_ANALYZER,
**options)
return props
def mapping(cls, idx, idx_type):
''' Load the mapping for the cytobands index.
seqid - chromosome
start - start position
stop - stop position
name - cytoband name
gieStain - Giemsa stain results. gneg, gpos50, gpos75, gpos25, gpos100, acen, gvar, stalk
'''
props = MappingProperties(idx_type)
props.add_property("seqid", "string", index="not_analyzed") \
.add_property("start", "integer") \
.add_property("stop", "integer") \
.add_property("name", "string", analyzer="full_name") \
.add_property("giestain", "string", index="not_analyzed")
''' create index and add mapping '''
load = Loader()
options = {"indexName": idx, "shards": 2}
load.mapping(props, idx_type, analyzer=Loader.KEYWORD_ANALYZER, **options)
return props
def _create_gene_mapping(self, **options):
''' Create the mapping for gene target index '''
props = MappingProperties("gene_target")
props.add_property("ensg", "string", index="not_analyzed") \
.add_property("name", "string", index="not_analyzed") \
.add_property("biotype", "string", index="not_analyzed") \
.add_property("strand", "string", index="no") \
.add_property("baitChr", "string", index="not_analyzed") \
.add_property("baitStart", "integer", index="not_analyzed") \
.add_property("baitEnd", "integer", index="not_analyzed") \
.add_property("baitID", "string", index="no") \
.add_property("baitName", "string", index="no") \
.add_property("oeChr", "string", index="not_analyzed") \
.add_property("oeStart", "integer", index="not_analyzed") \
.add_property("oeEnd", "integer", index="not_analyzed") \
.add_property("oeID", "string", index="no") \
.add_property("oeName", "string", index="no") \
.add_property("dist", "integer", index="not_analyzed")
meta = {"tissue_type": {}}
for tt in GeneTargetManager.tissue_types:
props.add_property(tt, "float")
meta["tissue_type"][tt] = "tissue_type"
self.mapping(props, idx_type='gene_target', meta=meta, **options)
def test_mapping_parent_child(self):
''' Test creating mapping with parent child relationship. '''
gene_mapping = MappingProperties("gene")
gene_mapping.add_property("symbol", "string", analyzer="full_name")
inta_mapping = MappingProperties("publication", "gene")
load = Loader()
idx = "test__mapping__"+SEARCH_SUFFIX
options = {"indexName": idx, "shards": 1}
requests.delete(ElasticSettings.url() + '/' + idx)
# add child mappings first
status = load.mapping(inta_mapping, "publication", analyzer=Loader.KEYWORD_ANALYZER, **options)
self.assertTrue(status, "mapping inteactions")
status = load.mapping(gene_mapping, "gene", analyzer=Loader.KEYWORD_ANALYZER, **options)
self.assertTrue(status, "mapping genes")
''' load docs and test has parent query'''
json_data = '{"index": {"_index": "%s", "_type": "gene", "_id" : "1"}}\n' % idx
json_data += json.dumps({"symbol": "PAX1"}) + '\n'
json_data += '{"index": {"_index": "%s", "_type": "publication", "_id" : "2", "parent": "1"}}\n' % idx
json_data += json.dumps({"pubmed": 1234}) + '\n'
Bulk.load(idx, '', json_data)
Search.index_refresh(idx)
query = ElasticQuery.has_parent('gene', Query.match('symbol', 'PAX1'))
elastic = Search(query, idx=idx, idx_type='publication', size=500)
docs = elastic.search().docs
self.assertEquals(len(docs), 1)
self.assertEquals(getattr(docs[0], 'pubmed'), 1234)
self.assertEquals(docs[0].parent(), '1')
self.assertRaises(QueryError, ElasticQuery.has_parent, 'gene', 'xxxxx')
''' test has child query '''
query = ElasticQuery.has_child('publication', Query.match('pubmed', 1234))
elastic = Search(query, idx=idx, idx_type='gene', size=500)
docs = elastic.search().docs
self.assertEquals(len(docs), 1)
self.assertEquals(getattr(docs[0], 'symbol'), 'PAX1')
self.assertEquals(docs[0].parent(), None)
requests.delete(ElasticSettings.url() + '/' + idx)
def mapping(cls, idx, idx_type):
''' Create the mapping for disease indexing '''
props = MappingProperties("disease")
props.add_property("name", "string") \
.add_property("code", "string") \
.add_property("description", "string", index="not_analyzed") \
.add_property("colour", "string", index="not_analyzed") \
.add_property("tier", "integer", index="not_analyzed") \
.add_property("suggest", "completion", analyzer="full_name")
tags = MappingProperties("tags")
tags.add_property("weight", "integer", index="not_analyzed")
props.add_properties(tags)
load = Loader()
options = {"indexName": idx, "shards": 1}
load.mapping(props, 'disease', analyzer=Loader.KEYWORD_ANALYZER, **options)
def _create_json_mapping(self, idx_type, mapping, **options):
''' Create the mapping for indexing '''
props = MappingProperties(idx_type)
props.mapping_properties[idx_type].update(mapping)
self.mapping(props, idx_type, **options)
def mapping(cls, idx, idx_type, meta):
''' Create the mapping for gwas/ic stats indexing '''
props = MappingProperties(idx_type)
props.add_property("seqid", "string")
props.add_property("position", "integer")
props.add_property("marker", "string", index="not_analyzed")
props.add_property("p_value", "double")
props.add_property("odds_ratio", "float")
props.add_property("lower_or", "float", index="no")
props.add_property("upper_or", "float", index="no")
props.add_property("raf", "float", index="no")
props.add_property("risk_allele", "string", index="no")
props.add_property("alt_allele", "string", index="no")
props.add_property("imputed", "byte", index="no")
load = Loader()
options = {"indexName": idx, "shards": 5}
load.mapping(props, idx_type, meta=meta, **options)
def _load_pathway_mappings(cls, section):
'''Function to load the elastic mappings'''
idx = section['index']
idx_type = section['index_type']
pathway_mapping = MappingProperties(idx_type)
pathway_mapping.add_property("pathway_name", "string")
pathway_mapping.add_property("pathway_url", "string")
pathway_mapping.add_property("gene_sets", "string")
pathway_mapping.add_property("source", "string")
pathway_mapping.add_property("is_public", "string")
load = Loader()
options = {"indexName": idx, "shards": 1}
status = load.mapping(pathway_mapping, idx_type, **options)
return status
def test_mapping_error(self):
self.assertRaises(LoaderError,
Loader().mapping, 'MappingProperties', '')
self.assertRaises(MappingError,
MappingProperties('').add_properties,
'MappingProperties')
def create_criteria_mapping(cls, idx, idx_type, test_mode=False):
''' function to create mapping for criteria indexes
@type idx: string
@param idx: name of the index
@type idx_type: string
@param idx_type: name of the idx type, each criteria is an index type
@type test_mode: string
@param test_mode: flag to create or not create the mapping
'''
logger.warning('Idx ' + idx)
logger.warning('Idx_type ' + idx_type)
''' Create the mapping for alias indexing '''
props = MappingProperties(idx_type)
props.add_property("score", "integer")
props.add_property("disease_tags", "string", index="not_analyzed")
props.add_property("qid", "string", index="not_analyzed")
(main_codes, other_codes) = CriteriaManager().get_available_diseases()
for disease in main_codes + other_codes:
criteria_tags = MappingProperties(disease)
criteria_tags.add_property("fid", "string", index="not_analyzed")
criteria_tags.add_property("fname", "string", index="not_analyzed")
fnotes = MappingProperties('fnotes')
fnotes.add_property('linkid', "string", index="not_analyzed")
fnotes.add_property('linkname', "string", index="not_analyzed")
fnotes.add_property('linkdata', "string", index="not_analyzed")
fnotes.add_property('linkvalue', "string", index="not_analyzed")
criteria_tags.add_properties(fnotes)
props.add_properties(criteria_tags)
''' create index and add mapping '''
load = Loader()
options = {"indexName": idx, "shards": 5}
'''add meta info'''
config = CriteriaManager.get_criteria_config()
idx_type_cfg = config[idx_type]
desc = idx_type_cfg['desc']
meta = {"desc": desc}
if not test_mode:
load.mapping(props, idx_type, meta=meta, analyzer=Loader.KEYWORD_ANALYZER, **options)
return props
def _get_nested_prop(cls, nested_name, prop_name):
org_props = MappingProperties(nested_name)
org_props.add_property(prop_name, "string", index="not_analyzed")
return org_props