Merge pull request #112 from bcgsc/release/v3.8.0_approved_levels_fro…

…m_IPR_A Release/v3.8.0 approved levels from ipr a
bcgsc · Apr 19, 2023 · 1494305 · 1494305
2 parents 6b15562 + 9faf8a8
commit 1494305
Show file tree

Hide file tree

Showing 5 changed files with 72 additions and 35 deletions.
diff --git a/ipr/constants.py b/ipr/constants.py
@@ -1,21 +1,6 @@
+DEFAULT_URL = 'https://iprstaging-api.bcgsc.ca/api'
 GERMLINE_BASE_TERMS = ('pharmacogenomic', 'cancer predisposition')  # based on graphkb.constants
 VARIANT_CLASSES = {'Variant', 'CategoryVariant', 'PositionalVariant', 'CatalogueVariant'}
 
-
-APPROVED_EVIDENCE_LEVELS = {
-    # sourceIds of levels by source name
-    'oncokb': ['1', 'r1'],
-    'profyle': ['t1'],
-    'cancer genome interpreter': [
-        'cpic guidelines',
-        'european leukemianet guidelines',
-        'fda guidelines',
-        'nccn guidelines',
-        'nccn/cap guidelines',
-    ],
-}
-
-DEFAULT_URL = 'https://iprstaging-api.bcgsc.ca/api'
-
 # all possible values for review status are: ['pending', 'not required', 'passed', 'failed', 'initial']
 FAILED_REVIEW_STATUS = 'failed'
diff --git a/ipr/content.spec.json b/ipr/content.spec.json
@@ -196,6 +196,13 @@
                             "integer",
                             "null"
                         ]
+                    },
+                    "size": {
+                        "description": "size of the copy change segment in the chromosome",
+                        "type": [
+                            "number",
+                            "null"
+                        ]
                     }
                 },
                 "required": [
@@ -511,6 +518,14 @@
                         "description": "the sequencing protocol used for ths sample",
                         "example": "DNA",
                         "type": "string"
+                    },
+                    "library": {
+                        "description": "Library id string",
+                        "example": "P01234",
+                        "type": [
+                            "string",
+                            "null"
+                        ]
                     }
                 },
                 "type": "object"
@@ -993,6 +1008,14 @@
                             "null"
                         ]
                     },
+                    "tumourAltCopies": {
+                        "description": "the number of gene copies in the tumour genome supporting the mutation",
+                        "example": 1,
+                        "type": [
+                            "integer",
+                            "null"
+                        ]
+                    },
                     "tumourAltCount": {
                         "description": "the number of alternate reads in the tumour genome supporting the mutation",
                         "example": 1,
@@ -1017,6 +1040,14 @@
                             "null"
                         ]
                     },
+                    "tumourRefCopies": {
+                        "description": "the number of wildtype gene copies in tumour genome",
+                        "example": 1,
+                        "type": [
+                            "integer",
+                            "null"
+                        ]
+                    },
                     "zygosity": {
                         "examples": [
                             "het",
@@ -1099,6 +1130,14 @@
                             "null"
                         ]
                     },
+                    "frame": {
+                        "description": "Optional descriptive string for in-frame vs frame-shift SV",
+                        "example": "frame-shift",
+                        "type": [
+                            "string",
+                            "null"
+                        ]
+                    },
                     "gene1": {
                         "description": "the 5' (n-terminal) gene name",
                         "example": "EWSR1",
@@ -1149,6 +1188,15 @@
                             "null"
                         ]
                     },
+                    "mavis_product_id": {
+                        "default": null,
+                        "description": "mavis product id / other tool description strings",
+                        "example": "MAVIS-2.5.35-RT3556GHTR",
+                        "type": [
+                            "string",
+                            "null"
+                        ]
+                    },
                     "svg": {
                         "description": "svg image file content for this SV",
                         "type": [
@@ -1162,6 +1210,22 @@
                             "string",
                             "null"
                         ]
+                    },
+                    "tumourAltCount": {
+                        "description": "the number of alternate reads in the tumour genome supporting the mutation",
+                        "example": 1,
+                        "type": [
+                            "integer",
+                            "null"
+                        ]
+                    },
+                    "tumourDepth": {
+                        "description": "the total number of reads at this position in the tumour genome",
+                        "example": 2,
+                        "type": [
+                            "integer",
+                            "null"
+                        ]
                     }
                 },
                 "required": [

diff --git a/ipr/inputs.py b/ipr/inputs.py
@@ -142,6 +142,8 @@
     # GERO-307 - tumourAltCount and tumourDepth are available but not rnaAltCount and rnaDepth
     'tumourAltCount',
     'tumourDepth',
+    'germline',
+    'mavis_product_id',
 ]
 
 

diff --git a/ipr/ipr.py b/ipr/ipr.py
@@ -8,7 +8,7 @@
 from graphkb.types import Record
 from typing import Dict, Iterable, List, Sequence, Set, Tuple
 
-from .constants import APPROVED_EVIDENCE_LEVELS, GERMLINE_BASE_TERMS, VARIANT_CLASSES
+from .constants import GERMLINE_BASE_TERMS, VARIANT_CLASSES
 from .types import GkbStatement, ImageDefinition, IprFusionVariant, IprGene, IprVariant, KbMatch
 from .util import convert_to_rid_set, find_variant, logger
 
@@ -23,20 +23,6 @@ def display_evidence_levels(statement: GkbStatement) -> str:
     return ';'.join(sorted(result))
 
 
-def get_approved_evidence_levels(graphkb_conn: GraphKBConnection) -> List[Record]:
-    filters = []
-    for source, names in APPROVED_EVIDENCE_LEVELS.items():
-        filters.append(
-            {
-                'AND': [
-                    {'source': {'target': 'Source', 'filters': {'name': source}}},
-                    {'name': names, 'operator': 'IN'},
-                ]
-            }
-        )
-    return graphkb_conn.query({'target': 'EvidenceLevel', 'filters': {'OR': filters}})
-
-
 def filter_structural_variants(
     structural_variants: List[IprFusionVariant],
     kb_matches: List[KbMatch],
@@ -86,7 +72,7 @@ def get_evidencelevel_mapping(graphkb_conn: GraphKBConnection) -> Dict[str, str]
 
     # Filter IPR EvidenceLevel and map each outgoing CrossReferenceOf to displayName
     ipr_source_rid = graphkb_conn.get_source("ipr")["@rid"]
-    ipr_evidence_levels = filter(lambda d: d["source"] == ipr_source_rid, evidence_levels)
+    ipr_evidence_levels = filter(lambda d: d.get("source") == ipr_source_rid, evidence_levels)
     cross_references_mapping: Dict[str, str] = dict()
     ipr_rids_to_displayname = dict()
     for level in ipr_evidence_levels:
@@ -141,9 +127,9 @@ def convert_statements_to_alterations(
         raise ValueError(f'failed to match disease ({disease_name}) to graphkb')
 
     rows = []
-
-    approved = convert_to_rid_set(get_approved_evidence_levels(graphkb_conn))
     ev_map = get_evidencelevel_mapping(graphkb_conn)
+    # GERO-318 - add all IPR-A evidence equivalents to the approvedTherapy flag
+    approved = set([ev for (ev, ipr) in ev_map.items() if ipr == 'IPR-A'])
 
     for statement in statements:
         variants = [c for c in statement['conditions'] if c['@class'] in VARIANT_CLASSES]

diff --git a/setup.cfg b/setup.cfg
@@ -18,7 +18,7 @@ known_standard_library = requests
 
 [metadata]
 name = ipr
-version = 3.7.1
+version = 3.8.0
 author_email = [email protected]
 author = ipr
 maintainer_email = [email protected]