Fix #4909 - Tavtigian Bayesian point sums

CHANGELOG.md

@@ -10,6 +10,7 @@ About changelog [here](https://keepachangelog.com/en/1.0.0/)
 - More SVI recommendation links on the ACMG page
 - IGV buttons for SMN CN page
 - Warnings on ACMG classifications for potentially conflicting classification pairs
+- ACMG Bayesian foundation point scale after Tavtigian for variant heat profile
 ### Changed
 - Variants query backend allows rank_score filtering
 - Added script to tabulate causatives clinical filter rank

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

@@ -113,15 +113,6 @@ <h4>Evidence of {{ category }}</h4>
   </div>
 {% endblock %}
 
-{% macro preview() %}
-  <div class="card panel-default">
-    <div class="card-body">
-      <span class="text-muted">Classification</span>
-      <h5>Likely Pathogenic</h5>
-    </div>
-  </div>
-{% endmacro %}
-
 {% block scripts %}
   {{ super() }}
 
@@ -173,10 +164,13 @@ <h5>Likely Pathogenic</h5>
         // add new selection
         $('#acmg-' + data.classification).addClass('btn-primary');
 
+        var temperature_span = document.getElementById("temperature_span");
+        temperature_span.innerHTML = 'Score ' + data.points + ' <span class="fa ' + data.temperature_icon + '"></span> ' + data.temperature + ' (' + data.point_classification + ')'
+        temperature_span.className = 'badge bg-' + data.temperature_class
+
         // Update any classification conflicts
         var conflicts_div = document.getElementById("conflicts_div");
         conflicts_div.innerHTML = data.conflicts.join("<br>");
-
       });
     }
   </script>

scout/server/blueprints/variant/views.py

@@ -26,7 +26,7 @@
 )
 from scout.server.extensions import loqusdb, store
 from scout.server.utils import institute_and_case, public_endpoint, templated
-from scout.utils.acmg import get_acmg, get_acmg_conflicts
+from scout.utils.acmg import get_acmg, get_acmg_conflicts, get_acmg_temperature
 from scout.utils.ensembl_rest_clients import EnsemblRestApiClient
 
 LOG = logging.getLogger(__name__)
@@ -347,8 +347,10 @@ def acmg():
     """Calculate an ACMG classification from submitted criteria."""
     criteria = request.args.getlist("criterion")
     classification = get_acmg(criteria)
+
+    acmg_bayesian = get_acmg_temperature(criteria)
     acmg_conflicts = get_acmg_conflicts(criteria)
-    return jsonify({"classification": classification, "conflicts": acmg_conflicts})
+    return jsonify({"classification": classification, "conflicts": acmg_conflicts, **acmg_bayesian})
 
 
 @variant_bp.route(

scout/utils/acmg.py

@@ -1,5 +1,9 @@
 # coding=UTF-8
 
+
+from typing import Optional
+
+from scout.constants import ACMG_COMPLETE_MAP
 from scout.constants.acmg import ACMG_POTENTIAL_CONFLICTS
 
 
@@ -161,38 +165,31 @@ def is_likely_benign(bs_terms, bp_terms):
     return False
 
 
-def get_acmg(acmg_terms):
-    """Use the algorithm described in ACMG paper to get a ACMG calssification
-
-    Modifying strength of a term is possible by adding a string describing its new level: "PP1_Strong" or
-    "PVS1_Moderate".
-
-    If no terms return None
-
-    Args:
-        acmg_terms(set(str)): A collection of prediction terms
-
-    Returns:
-        prediction(str): in ['uncertain_significance','benign','likely_benign',
-                             'likely_pathogenic','pathogenic']
-
+def get_acmg_criteria(acmg_terms: set) -> tuple:
     """
-    if not acmg_terms:
-        return None
-    prediction = "uncertain_significance"
-    # This variable indicates if Pathogenecity Very Strong exists
+    Given a set of ACMG evidence criteria terms, that may be strength modified with suffixes.
+    For each term,
+        first, Strength modified criteria suffixes should count towards their modified score.
+        then, we need to see if we match any of the two stand-alone terms. If so, set their respective booleans.
+        finally, check remaining prefixes if no suffix match or stand-alone criteria match
+
+    Return a tuple with
+        pvs: This variable indicates if Pathogenicity Very Strong exists
+        ps_terms: Collection of terms with Pathogenicity Strong
+        pm_terms: Collection of terms with Pathogenicity moderate
+        pp_terms: Collection of terms with Pathogenicity supporting
+        ba: This variable indicates if Benign impact stand-alone exists
+        bs_terms: Collection of terms with Benign evidence Strong
+        bp_terms: Collection of terms with supporting Benign evidence
+    """
+
     pvs = False
-    # Collection of terms with Pathogenecity Strong
     ps_terms = []
-    # Collection of terms with Pathogenecity moderate
     pm_terms = []
-    # Collection of terms with Pathogenecity supporting
     pp_terms = []
-    # This variable indicates if Benign impact stand-alone exists
+
     ba = False
-    # Collection of terms with Benign evidence Strong
     bs_terms = []
-    # Collection of terms with supporting Benign evidence
     bp_terms = []
 
     suffix_map = {
@@ -220,20 +217,47 @@ def get_acmg(acmg_terms):
                     continue
                 break
         else:
-            # Do we match any of the two standalone terms
             if term.startswith("PVS"):
                 pvs = True
             elif term.startswith("BA"):
                 ba = True
-            else:  # Check remaining prefixes if no suffix match or standalone criteria match
+            else:
                 for prefix, term_list in prefix_map.items():
                     if term.startswith(prefix):
                         term_list.append(term)
                         break
 
+    return (pvs, ps_terms, pm_terms, pp_terms, ba, bs_terms, bp_terms)
+
+
+def get_acmg(acmg_terms: set) -> Optional[str]:
+    """Use the algorithm described in ACMG paper (Richards 2015) to get a ACMG classification
+
+    Modifying strength of a term is possible by adding a string describing its new level: "PP1_Strong" or
+    "PVS1_Moderate".
+
+    BA is considered fully Stand Alone.
+
+    If no terms return None
+
+    Args:
+        acmg_terms(set(str)): A collection of prediction terms
+
+    Returns:
+        prediction(str): in ['uncertain_significance','benign','likely_benign',
+                             'likely_pathogenic','pathogenic']
+
+    """
+    if not acmg_terms:
+        return None
+
+    (pvs, ps_terms, pm_terms, pp_terms, ba, bs_terms, bp_terms) = get_acmg_criteria(acmg_terms)
+
     if ba:
         return "benign"
 
+    prediction = "uncertain_significance"
+
     pathogenic = is_pathogenic(pvs, ps_terms, pm_terms, pp_terms)
     likely_pathogenic = is_likely_pathogenic(pvs, ps_terms, pm_terms, pp_terms)
     benign = is_benign(ba, bs_terms)
@@ -255,6 +279,90 @@ def get_acmg(acmg_terms):
     return prediction
 
 
+def get_acmg_temperature(acmg_terms: set) -> Optional[dict]:
+    """
+    Use the algorithm described in Tavtigian 2020 to classifiy variants.
+
+    PVS 8 points, S 4, M 2, P 1.
+    This gives:
+
+    P > 10
+    LP 6 < p < 9
+    VUS 0 < p < 5
+    LB -1 < p < -6
+    B < -7
+
+    If no terms return None
+
+    Args:
+        acmg_terms(set(str)): A collection of prediction terms
+
+    Returns:
+        dict:
+            temperature:
+        (points, temperature, point_classification)
+
+    """
+    TEMPERATURE_STRINGS = {
+        -1: {"label": "B/LB", "color": "success", "icon": "fa-times"},
+        0: {"label": "Ice cold", "color": "info", "icon": "fa-icicles"},
+        1: {"label": "Cold", "color": "info", "icon": "fa-snowman"},
+        2: {"label": "Cold", "color": "info", "icon": "fa-snowflake"},
+        3: {"label": "Tepid", "color": "yellow", "icon": "fa-temperature-half"},
+        4: {"label": "Warm", "color": "orange", "icon": "fa-mug-hot"},
+        5: {"label": "Hot", "color": "red", "icon": "fa-pepper-hot"},
+        6: {"label": "LP/P", "color": "danger", "icon": "fa-stethoscope"},
+    }
+
+    if not acmg_terms:
+        return {}
+
+    (pvs, ps_terms, pm_terms, pp_terms, ba, bs_terms, bp_terms) = get_acmg_criteria(acmg_terms)
+
+    if ba:
+        points = -8
+    else:
+        points = (
+            8 * pvs
+            + 4 * len(ps_terms)
+            + 2 * len(pm_terms)
+            + len(pp_terms)
+            - 4 * len(bs_terms)
+            - len(bp_terms)
+        )
+
+    if points <= -7:
+        point_classification = "benign"
+        temperature_icon = TEMPERATURE_STRINGS[-1].get("icon")
+    elif points <= -1:
+        point_classification = "likely_benign"
+        temperature_icon = TEMPERATURE_STRINGS[-1].get("icon")
+    elif points <= 5:
+        point_classification = "uncertain_significance"
+    elif points <= 9:
+        point_classification = "likely_pathogenic"
+        temperature_icon = TEMPERATURE_STRINGS[6].get("icon")
+    elif points >= 10:
+        point_classification = "pathogenic"
+        temperature_icon = TEMPERATURE_STRINGS[6].get("icon")
+
+    temperature_class = ACMG_COMPLETE_MAP[point_classification].get("color")
+    temperature = ACMG_COMPLETE_MAP[point_classification].get("label")
+
+    if point_classification == "uncertain_significance":
+        temperature_class = TEMPERATURE_STRINGS[points].get("color")
+        temperature = TEMPERATURE_STRINGS[points].get("label")
+        temperature_icon = TEMPERATURE_STRINGS[points].get("icon")
+
+    return {
+        "points": points,
+        "temperature": temperature,
+        "temperature_class": temperature_class,
+        "temperature_icon": temperature_icon,
+        "point_classification": ACMG_COMPLETE_MAP[point_classification].get("short"),
+    }
+
+
 def get_acmg_conflicts(acmg_terms: set) -> list:
     """Check potential conflict paris, return list of reference strings."""
 

tests/utils/test_acmg.py

@@ -1,6 +1,7 @@
 from scout.utils.acmg import (
     get_acmg,
     get_acmg_conflicts,
+    get_acmg_temperature,
     is_benign,
     is_likely_benign,
     is_likely_pathogenic,
@@ -385,47 +386,47 @@ def test_get_acmg_no_terms():
 
 
 def test_get_acmg_pathogenic():
-    acmg_terms = ["PVS1", "PS1"]
+    acmg_terms = {"PVS1", "PS1"}
     res = get_acmg(acmg_terms)
     assert res == "pathogenic"
 
-    acmg_terms = ["PVS1", "PS1", "BS1"]
+    acmg_terms = {"PVS1", "PS1", "BS1"}
     res = get_acmg(acmg_terms)
     assert res == "pathogenic"
 
 
 def test_get_acmg_modifier():
-    acmg_terms = ["PVS1", "PS1"]
+    acmg_terms = {"PVS1", "PS1"}
     res = get_acmg(acmg_terms)
     assert res == "pathogenic"
 
-    acmg_terms = ["PVS1_Moderate", "PS1"]
+    acmg_terms = {"PVS1_Moderate", "PS1"}
     res = get_acmg(acmg_terms)
     assert res == "likely_pathogenic"
 
 
 def test_get_acmg_uncertain():
-    acmg_terms = ["PVS1"]
+    acmg_terms = {"PVS1"}
     res = get_acmg(acmg_terms)
     assert res == "uncertain_significance"
 
-    acmg_terms = ["PVS1", "PS1", "BS1", "BS2"]
+    acmg_terms = {"PVS1", "PS1", "BS1", "BS2"}
     res = get_acmg(acmg_terms)
     assert res == "uncertain_significance"
 
 
 def test_get_acmg_stand_alone_benign():
-    acmg_terms = ["PVS1", "PS1", "BA1"]
+    acmg_terms = {"PVS1", "PS1", "BA1"}
     res = get_acmg(acmg_terms)
     assert res == "benign"
 
-    bs_terms = ["BA1", "BS1", "BP1"]
+    acmg_terms = {"BA1", "BS1", "BP1"}
     res = get_acmg(acmg_terms)
     assert res == "benign"
 
 
 def test_acmg_modifier_on_both_benign_and_pathogenic():
-    acmg_terms = ["PS3_Moderate", "PP1_Moderate", "PP3", "BS1_Supporting"]
+    acmg_terms = {"PS3_Moderate", "PP1_Moderate", "PP3", "BS1_Supporting"}
     res = get_acmg(acmg_terms)
     assert res == "uncertain_significance"
 
@@ -440,3 +441,23 @@ def test_acmg_conflicts():
     acmg_terms = {"PVS1", "PM4"}
     conflicts = get_acmg_conflicts(acmg_terms)
     assert len(conflicts) == 1
+
+
+def test_acmg_temperature():
+    acmg_terms = {"PVS1", "PS1", "PP1", "BS1", "BS2"}
+    res = get_acmg_temperature(acmg_terms)
+    assert res["points"] == 5
+    assert res["temperature"] == "Hot"
+    assert res["point_classification"] == "VUS"
+
+    acmg_terms = {"PS3_Moderate", "PP1_Moderate", "PP3", "BS1_Supporting"}
+    res = get_acmg_temperature(acmg_terms)
+    assert res["points"] == 4
+    assert res["temperature"] == "Warm"
+    assert res["point_classification"] == "VUS"
+
+    acmg_terms = {"PVS1", "BS2", "BP1"}
+    res = get_acmg_temperature(acmg_terms)
+    assert res["points"] == 3
+    assert res["temperature"] == "Tepid"
+    assert res["point_classification"] == "VUS"