Fix #4442 - do not bring in ORPHA inheritance model to OMIM specific …

CHANGELOG.md

@@ -17,6 +17,7 @@ About changelog [here](https://keepachangelog.com/en/1.0.0/)
 ### Fixed
 - Removed log info showing hgnc IDs used in variantS search
 - Maintain Matchmaker Exchange and Beacon submission status when a case is re-uploaded
+- Inheritance mode from ORPHA should not be confounded with the OMIM inheritance model
 
 ### Added
 - New SO terms: `sequence_variant` and `coding_transcript_variant`

scout/server/blueprints/variant/controllers.py

@@ -232,7 +232,7 @@ def variant(
     # Update some case panels info from db and populate it on variant to avoid showing removed panels
     update_case_panels(store, case_obj)
     # The hierarchical call order is relevant: cases are used to populate variants
-    update_variant_case_panels(store, case_obj, variant_obj)
+    update_variant_case_panels(case_obj, variant_obj)
 
     associate_variant_genes_with_case_panels(store, variant_obj)
 
@@ -310,8 +310,8 @@ def variant(
 
     if variant_obj.get("genetic_models"):
         variant_models = set(model.split("_", 1)[0] for model in variant_obj["genetic_models"])
-        all_models = variant_obj.get("all_models", set())
-        variant_obj["is_matching_inheritance"] = set.intersection(variant_models, all_models)
+        omim_models = variant_obj.get("omim_models", set())
+        variant_obj["is_matching_inheritance"] = set.intersection(variant_models, omim_models)
 
     # Prepare classification information for visualisation
     classification = variant_obj.get("acmg_classification")

scout/server/blueprints/variant/templates/variant/gene_disease_relations.html

@@ -50,7 +50,7 @@
               {% if gene.common and gene.disease_terms %}
                 {% for disease_term in gene.disease_terms %}
                   {% if disease_term.source == 'ORPHA' %}
-                    <tr>
+                    <tr data-bs-toggle="tooltip" title="Some ORPHA disorders are phenotypic umbrella terms for multiple genetic entities. The inheritance models are in this case a set derived from all those entities, not necessarily the inheritance mode known for this gene.">
                       <td>
                           <a href="http://omim.org/entry/{{ gene.common.omim_id }}" rel="noopener" target="_blank">
                             {{ gene.common.hgnc_symbol }}

scout/server/blueprints/variant/utils.py

@@ -8,7 +8,7 @@
 LOG = logging.getLogger(__name__)
 
 
-def add_panel_specific_gene_info(panel_info):
+def add_panel_specific_gene_info(panel_info: List[dict]) -> dict:
     """Adds manually curated information from a gene panel to a gene
 
     The panel info is a list of dictionaries since there can be multiple infos about a panel.
@@ -61,13 +61,9 @@ def add_panel_specific_gene_info(panel_info):
     return panel_specific
 
 
-def update_representative_gene(variant_obj, variant_genes):
+def update_representative_gene(variant_obj: dict, variant_genes: List[dict]):
     """Set the gene with most severe consequence as being representative
     Used for display purposes
-
-    Args:
-        variant_obj(Variant): a variant object
-        variant_genes(list(Genes): a list of genes
     """
 
     if variant_genes:
@@ -93,7 +89,9 @@ def update_representative_gene(variant_obj, variant_genes):
         variant_obj["first_rep_gene"] = None
 
 
-def update_transcripts_information(variant_gene, hgnc_gene, variant_obj, genome_build=None):
+def update_transcripts_information(
+    variant_gene: dict, hgnc_gene: dict, variant_obj: dict, genome_build: Optional[str] = None
+):
     """Collect tx info from the hgnc gene and panels and update variant transcripts
 
     Since the hgnc information are continuously being updated we need to run this each time a
@@ -102,14 +100,7 @@ def update_transcripts_information(variant_gene, hgnc_gene, variant_obj, genome_
     This function will:
         - Add a dictionary with tx_id -> tx_info to the hgnc variant
         - Add information from the panel
-        - Adds a list of refseq transcripts
-
-    Args:
-        variant_gene(dict): the gene information from the variant
-        hgnc_gene(dict): the hgnc gene information
-        varaiant_obj(scout.models.Variant)
-        genome_build(str): genome build
-
+        - Adds a list of RefSeq transcripts
     """
     genome_build = genome_build or "37"
     disease_associated_no_version = variant_gene.get("disease_associated_no_version", set())
@@ -152,23 +143,18 @@ def update_transcripts_information(variant_gene, hgnc_gene, variant_obj, genome_
         if refseq_id in disease_associated_no_version:
             transcript["is_disease_associated"] = True
 
-        # Since a ensemble transcript can have multiple refseq identifiers we add all of
+        # Since an Ensembl transcript can have multiple RefSeq identifiers we add all of
         # those
         transcript["refseq_identifiers"] = hgnc_transcript.get("refseq_identifiers", [])
         transcript["change_str"] = transcript_str(transcript, hgnc_symbol)
 
 
-def update_variant_case_panels(store, case_obj, variant_obj):
+def update_variant_case_panels(case_obj: dict, variant_obj: dict):
     """Populate variant with case gene panels with info on e.g. if a panel was removed on variant_obj.
     Variant objects panels are only a list of matching panel names.
 
-    The case_obj should be up to date first. Call update_case_panels() as needed in context:
+    The case_obj should be up-to-date first. Call update_case_panels() as needed in context:
     to save some resources we do not call it here for each variant.
-
-    Args:
-        store(adapter.MongoAdapter)
-        case_obj(dict):     case_obj with updated panels - update_case_panels() first
-        variant_obj(dict)
     """
 
     variant_panel_names = variant_obj.get("panels") or []
@@ -181,55 +167,62 @@ def update_variant_case_panels(store, case_obj, variant_obj):
     variant_obj["case_panels"] = case_panel_objs
 
 
-def add_gene_info(store, variant_obj, gene_panels=None, genome_build=None):
+def get_extra_info(gene_panels: list) -> Dict[int, dict]:
+    """Parse out extra information from gene panels."""
+    extra_info = {}
+
+    for panel_obj in gene_panels:
+        for gene_info in panel_obj["genes"]:
+            hgnc_id = gene_info["hgnc_id"]
+            if hgnc_id not in extra_info:
+                extra_info[hgnc_id] = []
+
+            extra_info[hgnc_id].append(gene_info)
+    return extra_info
+
+
+def seed_genes_with_only_hgnc_id(variant_obj: dict):
+    """Seed genes structure for (STR) variants that have only hgnc_ids."""
+    if not variant_obj.get("genes") and variant_obj.get("hgnc_ids"):
+        variant_obj["genes"] = []
+        for hgnc_id in variant_obj.get("hgnc_ids"):
+            variant_gene = {"hgnc_id": hgnc_id}
+            variant_obj["genes"].append(variant_gene)
+
+
+def add_gene_info(
+    store: MongoAdapter,
+    variant_obj: dict,
+    gene_panels: Optional[List[dict]] = None,
+    genome_build: Optional[str] = None,
+):
     """Adds information to variant genes from hgnc genes and selected gene panels.
 
     Variants are annotated with gene and transcript information from VEP. In Scout the database
-    keeps updated and extended information about genes and transcript. This function will compliment
-     the VEP information with the updated database information.
-    Also there is sometimes additional information that are manually curated in the gene panels.
-    Only the selected panels passed to this function (typically default) are used.
+    keeps updated and extended information about genes and transcript. This function will complement
+    the VEP information with the updated database information.
+    Also there is sometimes additional information that is manually curated in the gene panels.
+    Only the selected panels passed to this function (typically the case default panels) are used.
     This information needs to be added to the variant before sending it to the template.
 
     This function will loop over all genes and add that extra information.
-
-    Args:
-        store(scout.adapter.MongoAdapter)
-        variant_obj(dict): A variant from the database
-        gene_panels(list(dict)): List of panels from database
-        genome_build(str)
-
-    Returns:
-        variant_obj
     """
     gene_panels = gene_panels or []
     genome_build = genome_build or "37"
 
-    # Add a variable that checks if there are any refseq transcripts
+    institute = store.institute(variant_obj["institute"])
 
-    # extra_info will hold information from gene panels
-    extra_info = {}
-    for panel_obj in gene_panels:
-        for gene_info in panel_obj["genes"]:
-            hgnc_id = gene_info["hgnc_id"]
-            if hgnc_id not in extra_info:
-                extra_info[hgnc_id] = []
+    extra_info = get_extra_info(gene_panels)
 
-            extra_info[hgnc_id].append(gene_info)
+    seed_genes_with_only_hgnc_id(variant_obj)
 
     # Loop over the genes in the variant object to add information
     # from hgnc_genes and panel genes to the variant object
+    # Add a variable that checks if there are any refseq transcripts
     variant_obj["has_refseq"] = False
     variant_obj["disease_associated_transcripts"] = []
     all_models = set()
 
-    # seed genes structure for (STR) variants that have only hgnc_ids
-    if not variant_obj.get("genes") and variant_obj.get("hgnc_ids"):
-        variant_obj["genes"] = []
-        for hgnc_id in variant_obj.get("hgnc_ids"):
-            variant_gene = {"hgnc_id": hgnc_id}
-            variant_obj["genes"].append(variant_gene)
-
     if variant_obj.get("genes"):
         for variant_gene in variant_obj["genes"]:
             hgnc_id = variant_gene["hgnc_id"]
@@ -251,8 +244,6 @@ def add_gene_info(store, variant_obj, gene_panels=None, genome_build=None):
             update_transcripts_information(variant_gene, hgnc_gene, variant_obj, genome_build)
 
             variant_gene["common"] = hgnc_gene
-
-            institute = store.institute(variant_obj["institute"])
             add_gene_links(variant_gene, genome_build, institute=institute)
 
             # Add disease associated transcripts from panel to variant
@@ -261,19 +252,39 @@ def add_gene_info(store, variant_obj, gene_panels=None, genome_build=None):
                 variant_obj["disease_associated_transcripts"].append(transcript_str)
 
             # Add the associated disease terms
-            disease_terms = store.disease_terms_by_gene(hgnc_id, filter_project={"inheritance": 1})
+            disease_terms = store.disease_terms_by_gene(
+                hgnc_id, filter_project={"inheritance": 1, "source": 1}
+            )
 
-            all_models = all_models.union(set(variant_gene["manual_inheritance"]))
-            omim_models = set()
-            for disease_term in disease_terms:
-                omim_models.update(disease_term.get("inheritance", []))
-            variant_gene["omim_inheritance"] = list(omim_models)
+            all_models.update(set(variant_gene["manual_inheritance"]))
 
-            all_models = all_models.union(omim_models)
+            update_inheritance_model(variant_gene, all_models, disease_terms)
 
     variant_obj["all_models"] = all_models
 
 
+def update_inheritance_model(variant_gene: dict, all_models: set, disease_terms: list):
+    """Update OMIM inheritance model for variant gene - and update the all models
+    variable to contain all inheritance models suggested for the gene/disorder.
+
+    ORPHA disorders can be more of the umbrella kind, where many genes and inheritance
+    models are implied. Those models are still added to all_models for the variant, but not to the list
+    of OMIM inheritance models for the particular gene.
+    """
+    inheritance_models = set()
+    omim_inheritance_models = set()
+
+    for disease_term in disease_terms:
+        inheritance_models.update(disease_term.get("inheritance", []))
+
+        if disease_term.get("source") == "OMIM":
+            omim_inheritance_models.update(inheritance_models)
+
+    variant_gene["omim_inheritance"] = list(omim_inheritance_models)
+
+    all_models.update(inheritance_models)
+
+
 def predictions(genes):
     """Adds information from variant specific genes to display.
 
@@ -542,18 +553,12 @@ def end_position(variant_obj):
     return variant_obj["position"] + (num_bases - 1)
 
 
-def default_panels(store, case_obj):
+def default_panels(store: MongoAdapter, case_obj: dict) -> List[dict]:
     """Return the panels that are decided to be default for a case.
 
-    Check what panels that are default, fetch those and add them to a list.
-
-    Args:
-        store(scout.adapter.MongoAdapter)
-        case_obj(scout.models.Case)
-
-    Returns:
-        default_panels(list(dict))
+    Check what panels that are default, fetch those and return them in a list.
 
+    case_obj is a dict after scout.models.Case.
     """
     default_panels = []
     # Add default panel information to variant

scout/server/blueprints/variants/controllers.py

@@ -1,7 +1,6 @@
 import decimal
 import logging
 import re
-from datetime import date
 from typing import Any, Dict
 
 import bson
@@ -149,7 +148,7 @@ def variants(
         if case_obj.get("group"):
             variant_obj["group_assessments"] = _get_group_assessments(store, case_obj, variant_obj)
 
-        update_variant_case_panels(store, case_obj, variant_obj)
+        update_variant_case_panels(case_obj, variant_obj)
 
         variants.append(
             parse_variant(