scout-browser 4.86__py3-none-any.whl → 4.86.1__py3-none-any.whl

scout/__version__.py

@@ -1 +1 @@
-__version__ = "4.86"
+__version__ = "4.86.1"

scout/adapter/mongo/hgnc.py

@@ -7,6 +7,7 @@ from pymongo.errors import BulkWriteError, DuplicateKeyError
 from scout.exceptions import IntegrityError
 
 LOG = logging.getLogger(__name__)
+QUERY_FIELD_EXISTS = {"$exists": True, "$ne": None}
 
 
 class GeneHandler(object):
@@ -364,9 +365,13 @@ class GeneHandler(object):
         Returns:
             mapping(dict): {"ENSG00000121410": 5, ...}
         """
-        pipeline = [{"$group": {"_id": {"ensembl_id": "$ensembl_id", "hgnc_id": "$hgnc_id"}}}]
-        result = self.hgnc_collection.aggregate(pipeline)
-        mapping = {res["_id"]["ensembl_id"]: res["_id"]["hgnc_id"] for res in result}
+        query = {
+            "ensembl_id": QUERY_FIELD_EXISTS,
+            "hgnc_id": QUERY_FIELD_EXISTS,
+        }
+        project = {"ensembl_id": 1, "hgnc_id": 1}
+        result = self.hgnc_collection.find(query, project)
+        mapping = {res["ensembl_id"]: res["hgnc_id"] for res in result}
         return mapping
 
     def hgnc_symbol_ensembl_id_mapping(self) -> Dict[str, str]:
@@ -375,18 +380,13 @@ class GeneHandler(object):
         Returns:
            mapping(dict): {"A1BG": "ENSG00000121410".}
         """
-        pipeline = [
-            {
-                "$group": {
-                    "_id": {
-                        "hgnc_symbol": "$hgnc_symbol",
-                        "ensembl_id": "$ensembl_id",
-                    }
-                }
-            }
-        ]
-        result = self.hgnc_collection.aggregate(pipeline)
-        mapping = {res["_id"]["hgnc_symbol"]: res["_id"]["ensembl_id"] for res in result}
+        query = {
+            "ensembl_id": QUERY_FIELD_EXISTS,
+            "hgnc_symbol": QUERY_FIELD_EXISTS,
+        }
+        project = {"ensembl_id": 1, "hgnc_symbol": 1}
+        result = self.hgnc_collection.find(query, project)
+        mapping = {res["hgnc_symbol"]: res["ensembl_id"] for res in result}
         return mapping
 
     def ensembl_genes(self, build=None, add_transcripts=False, id_transcripts=False):

scout/load/panel.py

@@ -117,27 +117,33 @@ def _panelapp_panel_ids() -> List[str]:
     return [panel_info["Panel_Id"] for panel_info in json_lines.get("result", [])]
 
 
-def _parse_panelapp_panel(adapter, panel_id, institute, confidence):
+def _parse_panelapp_panel(
+    panel_id: str,
+    institute: str,
+    confidence: str,
+    ensembl_id_to_hgnc_id_map: Dict[str, int],
+    hgnc_symbol_to_ensembl_id_map: Dict[str, str],
+):
     """fetch and parse lines from a PanelApp panel, given its ID
 
     Args:
         adapter(scout.adapter.MongoAdapter)
         panel_id(str): The panel app panel id
         confidence(str enum green|amber|red): traffic light-style PanelApp level of confidence
+        ensembl_to_hgnc_map: dict[str, int]
+        hgnc_to_ensembl_map: [int, str]
+
 
     Returns:
         parsed_panel(dict). Example:
             {'version': 3.3, 'date': datetime.datetime(2023, 1, 31, 16, 43, 37, 521719), 'display_name': 'Diabetes - neonatal onset - [GREEN]', 'institute': 'cust000', 'panel_type': 'clinical', 'genes': [list of genes], 'panel_id': '55a9041e22c1fc6711b0c6c0'}
 
     """
-    ensembl_gene_hgnc_id_map: Dict[str, int] = adapter.ensembl_to_hgnc_id_mapping()
-    hgnc_symbol_ensembl_gene_map: Dict[str, str] = adapter.hgnc_symbol_ensembl_id_mapping()
-
     json_lines = fetch_resource(PANELAPP_BASE_URL.format("get_panel") + panel_id, json=True)
     parsed_panel = parse_panel_app_panel(
         panel_info=json_lines["result"],
-        ensembl_gene_hgnc_id_map=ensembl_gene_hgnc_id_map,
-        hgnc_symbol_ensembl_gene_map=hgnc_symbol_ensembl_gene_map,
+        ensembl_gene_hgnc_id_map=ensembl_id_to_hgnc_id_map,
+        hgnc_symbol_ensembl_gene_map=hgnc_symbol_to_ensembl_id_map,
         institute=institute,
         confidence=confidence,
     )
@@ -166,8 +172,17 @@ def load_panelapp_panel(adapter, panel_id=None, institute="cust000", confidence=
         LOG.info("Fetching all panel app panels")
         panel_ids: List[str] = _panelapp_panel_ids()
 
+    ensembl_id_to_hgnc_id_map: Dict[str, int] = adapter.ensembl_to_hgnc_id_mapping()
+    hgnc_symbol_to_ensembl_id_map: Dict[int, str] = adapter.hgnc_symbol_ensembl_id_mapping()
+
     for _ in panel_ids:
-        parsed_panel = _parse_panelapp_panel(adapter, _, institute, confidence)
+        parsed_panel = _parse_panelapp_panel(
+            panel_id=_,
+            institute=institute,
+            confidence=confidence,
+            ensembl_id_to_hgnc_id_map=ensembl_id_to_hgnc_id_map,
+            hgnc_symbol_to_ensembl_id_map=hgnc_symbol_to_ensembl_id_map,
+        )
 
         if len(parsed_panel["genes"]) == 0:
             LOG.warning("Panel %s is missing genes. Skipping.", parsed_panel["display_name"])
@@ -201,9 +216,19 @@ def load_panelapp_green_panel(adapter, institute, force):
     }
     genes = set()  # avoid duplicate genes from different panels
     # Loop over all PanelApp panels
+
+    ensembl_id_to_hgnc_id_map: Dict[str, int] = adapter.ensembl_to_hgnc_id_mapping()
+    hgnc_symbol_to_ensembl_id_map: Dict[int, str] = adapter.hgnc_symbol_ensembl_id_mapping()
+
     for _ in panel_ids:
         # And collect their green genes
-        parsed_panel = _parse_panelapp_panel(adapter, _, institute, "green")
+        parsed_panel = _parse_panelapp_panel(
+            panel_id=_,
+            institute=institute,
+            confidence="green",
+            ensembl_id_to_hgnc_id_map=ensembl_id_to_hgnc_id_map,
+            hgnc_symbol_to_ensembl_id_map=hgnc_symbol_to_ensembl_id_map,
+        )
         genes.update({(gene["hgnc_id"], gene["hgnc_symbol"]) for gene in parsed_panel.get("genes")})
 
     green_panel["genes"] = [{"hgnc_id": tup[0], "hgnc_symbol": tup[1]} for tup in genes]

{scout_browser-4.86.dist-info → scout_browser-4.86.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: scout-browser
-Version: 4.86
+Version: 4.86.1
 Summary: Clinical DNA variant visualizer and browser.
 Home-page: https://github.com/Clinical-Genomics/scout
 Author: Måns Magnusson

{scout_browser-4.86.dist-info → scout_browser-4.86.1.dist-info}/RECORD

@@ -1,5 +1,5 @@
 scout/__init__.py,sha256=Z4liXvmEcLkC67ElsWvYHfemPKdWgWI5O6MB6XlDM8M,232
-scout/__version__.py,sha256=zYKQx08a1gvbjSpG6DiyHGDUoAWcbw-0P5ZAvghe_Lg,21
+scout/__version__.py,sha256=Bv6pkrw-9AgvxM3NeIK6_fIwVSVxwIL91re4O83Ec0k,23
 scout/adapter/__init__.py,sha256=-iX_hx2NI1EMAqX0pMd5_90Nnd9uvIMxv9EbefYBzsc,86
 scout/adapter/client.py,sha256=IuajRsEwTG41ZP14X09Q1Cj94zIgmIvUtlXfcAFn0EA,1513
 scout/adapter/mongo/__init__.py,sha256=NdHYCUXWUAuX5cUS3-6HCws2hW9uoGep8i0SC-oJd3k,31
@@ -13,7 +13,7 @@ scout/adapter/mongo/cytoband.py,sha256=2NoIAoQ7RPqcipiSymD2O9qH8lXscP3T5ejquOrD7
 scout/adapter/mongo/disease_terms.py,sha256=4qcQRLuP5xn_RkB2RlNtwCrONmDuWwdlngZrj87BUEo,5637
 scout/adapter/mongo/event.py,sha256=NnURFhS5odOXk7So_gGMHmALPAJM0PWzESI4FC06azE,19842
 scout/adapter/mongo/filter.py,sha256=vzBJS_6vq2z76SH8PszdRR-MNuZ7_RA4VJvnkNYUs7s,8937
-scout/adapter/mongo/hgnc.py,sha256=Zcnf_kKJqrJAf62tUFODhL8OQUb695XnL_rzk4D9nes,16411
+scout/adapter/mongo/hgnc.py,sha256=9jbrkaTtlnnuE5gsSM8ylb8Kero4lC2zCWr-90QlyAY,16420
 scout/adapter/mongo/hpo.py,sha256=ps2Pix3sGeea7DenboZVP5j7Vh9WAgX3-sgpbAj0Eck,6871
 scout/adapter/mongo/index.py,sha256=TZPVv6xBWsEtZpBUWp5uhjuPjQMfVss4lZZju49g1Lw,3577
 scout/adapter/mongo/institute.py,sha256=GQmMKb_VzsvWwadp6gXHlSSpcG5ovF1ozdTzvpTFLpA,7025
@@ -370,7 +370,7 @@ scout/load/exon.py,sha256=ONaWJ9EfbhGV8KGXxwlq_jYCoU-JkBpv4r9zDaCjtls,2070
 scout/load/hgnc_gene.py,sha256=GcSQ01CioYIshAzXAv4msIGDPjTjGBBPBnLX2GvaZJ4,4507
 scout/load/hpo.py,sha256=ZWYjV9j6OfpI-ZZuhTQVEs_xgLa7ST2KuvYLW6ku0PY,2474
 scout/load/institute.py,sha256=etmZ0CZRHOVi3f9R-wn5BK1pZPhtZwq2QYbh2RXXxDc,625
-scout/load/panel.py,sha256=JQqkyUdpvhFnW81n8Sq-iPahDcm1hGBzLfrwUcVqt4k,9446
+scout/load/panel.py,sha256=Zxv-ZlRyk7ZWKy1p0Cod-ATh7TzKqeMbiLu5I6-VEzs,10246
 scout/load/report.py,sha256=TJvblks6Dp5-UDflWRTa8b3fe-A4bT4QnWT2yZLhfLA,820
 scout/load/setup.py,sha256=0qQgfCr53oHL626hG5y0o5vwRnJ4X0IMaTFvMBMRNQE,8183
 scout/load/transcript.py,sha256=mk0Da7GWwqLVRp2xRD9mlzqaomH9R-TL2aKCJx4fEA8,4632
@@ -671,9 +671,9 @@ scout/utils/md5.py,sha256=KkgdxOf7xbF9AF40ZjQKCgWaxFWJ9tp9RKjd8SU6IoA,649
 scout/utils/scout_requests.py,sha256=JjHOJW1XmenG05mNQ33kvOKq_IicveIfQMcPZeRcQdo,12856
 scout/utils/sort.py,sha256=1AcbeZ6vdt_UXM3BLDBa3aQmN4qxrqtskxwD19oBhvw,756
 scout/utils/track_resources.py,sha256=eUjSEe-Ff8BIb4BHPC_COkJocQO2PaWueiPz1GAuiwY,2614
-scout_browser-4.86.dist-info/LICENSE,sha256=TM1Y9Cqbwk55JVfxD-_bpGLtZQAeN9RovQlqHK6eOTY,1485
-scout_browser-4.86.dist-info/METADATA,sha256=LIrl5lLOrztC7-iyc2X-AAASuh6TvvWjA8Tvq09N4xg,14260
-scout_browser-4.86.dist-info/WHEEL,sha256=eOLhNAGa2EW3wWl_TU484h7q1UNgy0JXjjoqKoxAAQc,92
-scout_browser-4.86.dist-info/entry_points.txt,sha256=q_mxFwbMFTwXRDDIRVcqKram2ubMVmvs3CSNvZri1nY,45
-scout_browser-4.86.dist-info/top_level.txt,sha256=qM75h71bztMaLYsxn1up4c_n2rjc_ZnyaW6Q0K5uOXc,6
-scout_browser-4.86.dist-info/RECORD,,
+scout_browser-4.86.1.dist-info/LICENSE,sha256=TM1Y9Cqbwk55JVfxD-_bpGLtZQAeN9RovQlqHK6eOTY,1485
+scout_browser-4.86.1.dist-info/METADATA,sha256=mQw8mIrZPRWvJexXxlV4DL3-tNw3gL_XJQy38ROnv20,14262
+scout_browser-4.86.1.dist-info/WHEEL,sha256=eOLhNAGa2EW3wWl_TU484h7q1UNgy0JXjjoqKoxAAQc,92
+scout_browser-4.86.1.dist-info/entry_points.txt,sha256=q_mxFwbMFTwXRDDIRVcqKram2ubMVmvs3CSNvZri1nY,45
+scout_browser-4.86.1.dist-info/top_level.txt,sha256=qM75h71bztMaLYsxn1up4c_n2rjc_ZnyaW6Q0K5uOXc,6
+scout_browser-4.86.1.dist-info/RECORD,,