Phenotype matching (#426)

* Adds explicit phenotype matching step, using gen~phen db

* always do a strict comparison, optionally do semantic comparison

* Bump version: 5.1.4 → 5.2.0

* Co-authored-by: cassimons <[email protected]>

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 5.1.4
+current_version = 5.2.0
 commit = True
 tag = False
 

.github/workflows/index_page_builder.yaml

@@ -16,7 +16,7 @@ jobs:
     runs-on: ubuntu-latest
 
     env:
-      VERSION: 5.1.4
+      VERSION: 5.2.0
 
     steps:
 

helpers/stratify_solved_cases.py

@@ -9,7 +9,6 @@
 from sys import argv
 
 import toml
-
 from metamist.graphql import gql, query
 
 PROJECT_QUERY = gql(

pyproject.toml

@@ -55,5 +55,5 @@ quote-style = "single"
 section-order = ["future", "standard-library", "third-party", "hail", "cpg", "first-party", "local-folder"]
 
 [tool.ruff.lint.isort.sections]
-cpg = ["metamist"]
+cpg = ["metamist", "talos"]
 hail = ["hail", "hailtop"]

requirements.txt

@@ -12,5 +12,6 @@ peds>=1.2.0
 pydantic>=2.5.2
 pyspark>=3.5.1
 requests>=2.31.0
+semsimian>=0.2.15
 tabulate>=0.8.9
 toml==0.10.2

setup.py

@@ -27,7 +27,7 @@ def read_reqs(filename: str) -> list[str]:
     name='talos',
     description='Centre for Population Genomics Variant Prioritisation',
     long_description=readme,
-    version='5.1.4',
+    version='5.2.0',
     author='Matthew Welland, CPG',
     author_email='[email protected], [email protected]',
     package_data={'talos': ['templates/*.jinja', 'example_config.toml']},
@@ -65,12 +65,19 @@ def read_reqs(filename: str) -> list[str]:
             'GeneratePanelData = talos.GeneratePanelData:cli_main',
             # query PanelApp for those selected panels
             'QueryPanelapp = talos.QueryPanelapp:cli_main',
+            # use API queries to find the gene symbol for each gene ID
+            'FindGeneSymbolMap = talos.FindGeneSymbolMap:cli_main',
+            # # TODO - this thing just doesn't work in its current form. Does not scale.
+            # # match participant HPOs to gene HPOs for prioritisation
+            # # 'MatchGenesToPhenotypes = talos.MatchGenesToPhenotypes:cli_main',  # noqa: ERA001
             # Filter and label a small-variant MatrixTable
             'RunHailFiltering = talos.RunHailFiltering:cli_main',
             # Filter and label a SV MatrixTable
             'RunHailFilteringSV = talos.RunHailFilteringSV:cli_main',
             # Run each of the category-labelled variants through MOI filters
             'ValidateMOI = talos.ValidateMOI:cli_main',
+            # catch variants which have strong phenotypic matches
+            'HPOFlagging = talos.HPOFlagging:cli_main',
             # CPG internal (?), publish those results as an HTML report
             'CreateTalosHTML = talos.CreateTalosHTML:cli_main',
             # CPG internal (?), generate a file for ingestion by Seqr

src/talos/BuildReportIndexPage.py

@@ -7,13 +7,12 @@
 
 import re
 from dataclasses import dataclass
+from functools import lru_cache
 from pathlib import Path
 from typing import Any
-from functools import lru_cache
 
 import jinja2
 from cloudpathlib.anypath import to_anypath
-
 from metamist.graphql import gql, query
 
 from talos.static_values import get_logger

src/talos/CreateTalosHTML.py

@@ -469,6 +469,9 @@ def __init__(
         self.genotypes = report_variant.genotypes
         self.sample = sample
         self.ext_labels = ext_labels
+        # add the phenotype match date and HPO term id/labels
+        self.phenotype_match_date = report_variant.date_of_phenotype_match
+        self.phenotype_matches = report_variant.phenotype_labels
 
         # List of (gene_id, symbol)
         self.genes: list[tuple[str, str]] = []

src/talos/FindGeneSymbolMap.py

@@ -0,0 +1,87 @@
+"""
+Takes the PanelApp data (the full region of interest)
+produce an output of the {gene ID: gene symbol} for all included genes
+write this to a file
+
+This has been extracted to a separate script to allow for parallelisation
+and to ensure that we can separately repeat this one step in isolation
+if we run into API/throttling issues
+
+This may  be integrated into the HPO~Phenotype matching script if it runs consistently enough
+"""
+
+import asyncio
+import json
+from argparse import ArgumentParser
+
+from aiohttp import ClientSession
+
+from talos.config import config_retrieve
+from talos.models import PanelApp
+from talos.utils import chunks, read_json_from_path
+
+ENSEMBL_REST_API = 'http://rest.ensembl.org'
+
+
+async def match_ensgs_to_symbols(genes: list[str], session: ClientSession) -> dict[str, str]:
+    data_payload = json.dumps({'ids': genes})
+    r = await session.request(
+        method='POST',
+        url=f'{ENSEMBL_REST_API}/lookup/id',
+        headers={'Accept': 'application/json', 'Content-Type': 'application/json'},
+        data=data_payload,
+    )
+    r.raise_for_status()
+    json_reponse = await r.json()
+    # match symbol to the ENSG (or Unknown if the key is missing, or has a None value)
+    return {value.get('display_name'): key for key, value in json_reponse.items() if value}
+
+
+async def match_symbol_to_ensg(gene_symbol: str, session: ClientSession) -> tuple[str, str]:
+    r = await session.request(
+        method='GET',
+        url=f'{ENSEMBL_REST_API}/lookup/id/{gene_symbol}',
+        headers={'Content-Type': 'application/json'},
+    )
+    r.raise_for_status()
+    json_reponse = await r.json()
+    return gene_symbol, json_reponse['display_name']
+
+
+def cli_main():
+    parser = ArgumentParser()
+    parser.add_argument('--panelapp', help='Path to the PanelApp results file', required=True)
+    parser.add_argument('--out_path', help='where to write the output (.json)', required=True)
+    args = parser.parse_args()
+
+    main(args.panelapp, out_path=args.out_path)
+
+
+def main(panelapp_path: str, out_path: str):
+    """
+
+    Args:
+        panelapp_path (str): path to the PanelApp model JSON
+        out_path ():
+    """
+    panelapp_object = read_json_from_path(read_path=panelapp_path, return_model=PanelApp)
+    # confirm for mypy
+    assert isinstance(panelapp_object, PanelApp)
+    genes = list(panelapp_object.genes.keys())
+    ensg_to_symbol_mapping = asyncio.run(get_symbols_for_ensgs(genes))
+
+    with open(out_path, 'w') as f:
+        json.dump(ensg_to_symbol_mapping, f, indent=4)
+
+
+async def get_symbols_for_ensgs(genes: list[str]) -> dict[str, str]:
+    chunksize = config_retrieve(['FindGeneSymbolMap', 'chunk_size'], 800)
+    all_results: dict[str, str] = {}
+    async with ClientSession() as session:
+        for chunk in chunks(genes, chunksize):
+            all_results |= await match_ensgs_to_symbols(chunk, session=session)
+    return all_results
+
+
+if __name__ == '__main__':
+    cli_main()

src/talos/GeneratePanelData.py

@@ -16,14 +16,18 @@
 from peds import open_ped
 
 from talos.config import config_retrieve
-from talos.models import ParticipantHPOPanels, PhenotypeMatchedPanels
+from talos.models import ParticipantHPOPanels, PhenoPacketHpo, PhenotypeMatchedPanels
 from talos.static_values import get_logger
 
 HPO_RE = re.compile(r'HP:[0-9]+')
 MAX_DEPTH = 3
 
 PANELAPP_HARD_CODED_DEFAULT = 'https://panelapp.agha.umccr.org/api/v1/panels'
-PANELS_ENDPOINT = config_retrieve(['GeneratePanelData', 'panelapp'], PANELAPP_HARD_CODED_DEFAULT)
+try:
+    PANELS_ENDPOINT = config_retrieve(['GeneratePanelData', 'panelapp'], PANELAPP_HARD_CODED_DEFAULT)
+except KeyError:
+    get_logger(__file__).warning('Config environment variable TALOS_CONFIG not set, falling back to Aussie PanelApp')
+    PANELS_ENDPOINT = PANELAPP_HARD_CODED_DEFAULT
 
 
 def get_json_response(url: str) -> dict:
@@ -102,7 +106,7 @@ def get_participant_hpos(pedigree: str) -> tuple[PhenotypeMatchedPanels, set[str
             hpo_dict.samples[internal_id] = ParticipantHPOPanels(
                 external_id=external_id,
                 family_id=family_id,
-                hpo_terms=[{'id': hpo, 'label': ''} for hpo in member_data[1:]],
+                hpo_terms=[PhenoPacketHpo(id=hpo, label='') for hpo in member_data[1:]],
                 panels={137},
             )