Merge pull request #104 from cancervariants/gene-normalizer

Gene normalizer fix

Author: korikuzma

.gitignore

@@ -37,4 +37,5 @@ pyproject.toml
 # Zip
 *.zip
 
-build/
+build/
+dynamodb_local_latest/

variant/__init__.py

@@ -4,8 +4,9 @@
 import logging
 from ftplib import FTP
 from os import environ
+from gene.query import QueryHandler as GeneQueryHandler
 
-__version__ = "0.2.0"
+__version__ = "0.2.1"
 
 APP_ROOT = Path(__file__).resolve().parents[0]
 
@@ -16,6 +17,8 @@
 else:
     LOG_FN = f'{APP_ROOT}/variant/variant.log'
 
+GENE_NORMALIZER = GeneQueryHandler()
+
 logger = logging.getLogger('variant')
 if Path(LOG_FN).exists():
     fhandler = logging.FileHandler(filename=LOG_FN)

variant/data_sources/transcript_mappings.py

@@ -3,7 +3,6 @@
 from typing import Dict, List, Optional
 from variant.schemas.validation_response_schema import LookupType
 from variant import TRANSCRIPT_MAPPINGS_PATH, REFSEQ_GENE_SYMBOL_PATH
-from gene.query import QueryHandler as GeneQueryHandler
 
 
 class TranscriptMappings:
@@ -39,8 +38,6 @@ def __init__(self, transcript_file_path=TRANSCRIPT_MAPPINGS_PATH,
         self.refseq_lrg_for_gene_symbol: Dict[str, List[str]] = {}
         self.refseq_lrg_to_gene_symbol: Dict[str, str] = {}
 
-        self.gene_query_handler = GeneQueryHandler()
-
         self._load_transcript_mappings_data(transcript_file_path)
         self._load_refseq_gene_symbol_data(refseq_file_path)
 

variant/normalize.py

@@ -1,10 +1,10 @@
 """Module for Variant Normalization."""
 from typing import Tuple, Optional
+from variant import GENE_NORMALIZER
 from variant.schemas.token_response_schema import PolypeptideSequenceVariant,\
     SingleNucleotideVariant, DelIns, AminoAcidDelInsToken
 from variant.schemas.ga4gh_vod import Gene, VariationDescriptor, GeneDescriptor
 from variant.data_sources import SeqRepoAccess
-from gene.query import QueryHandler as GeneQueryHandler
 from urllib.parse import quote
 from variant import logger
 
@@ -14,7 +14,6 @@ class Normalize:
 
     def __init__(self):
         """Initialize Normalize class."""
-        self.gene_query_handler = GeneQueryHandler()
         self.seqrepo_access = SeqRepoAccess()
         self.warnings = list()
 
@@ -84,8 +83,7 @@ def get_gene_descriptor(self, gene_token):
             gene-normalizer.
         """
         gene_symbol = gene_token.matched_value
-        response = self.gene_query_handler.search_sources(gene_symbol,
-                                                          incl='hgnc')
+        response = GENE_NORMALIZER.search_sources(gene_symbol, incl='hgnc')
         if response['source_matches'][0]['records']:
             record = response['source_matches'][0]['records'][0]
             record_location = record.locations[0] if record.locations else None

variant/validators/genomic_base.py

@@ -1,5 +1,5 @@
 """Module for Genomic Validation methods."""
-from gene.query import QueryHandler as GeneQueryHandler
+from variant import GENE_NORMALIZER
 from ga4gh.vrs.dataproxy import SeqRepoDataProxy
 import logging
 
@@ -29,9 +29,8 @@ def get_nc_accessions(self, classification):
             gene_tokens = [t for t in classification.all_tokens
                            if t.token_type == 'GeneSymbol']
             if gene_tokens and len(gene_tokens) == 1:
-                gene_query_handler = GeneQueryHandler()
-                resp = gene_query_handler.search_sources(gene_tokens[0].token,
-                                                         incl='hgnc')
+                resp = GENE_NORMALIZER.search_sources(gene_tokens[0].token,
+                                                      incl='hgnc')
                 if resp['source_matches'][0]['records']:
                     record = resp['source_matches'][0]['records'][0]
                     loc = record.locations[0] if record.locations else None