feat: changes to PICS credible sets (OUT_OF_SAMPLE_LD QC flag and cap…

…ital PICS) (opentargets#910) * feat: add OUT_OF_SAMPLE_LD QC flag to PICS credible sets * feat: change pics finemapping method to PICS * test: change pics to PICS in test data * fix: flag studies without sumstats without relying on hasSumstats column * fix: flag studies without sumstats without using update_quality_flag function
thehyve · Nov 12, 2024 · e5b3c9e · e5b3c9e
1 parent 10b4be0
commit e5b3c9e
Show file tree

Hide file tree

Showing 4 changed files with 32 additions and 27 deletions.
diff --git a/src/gentropy/dataset/study_locus.py b/src/gentropy/dataset/study_locus.py
@@ -148,7 +148,7 @@ class FinemappingMethod(Enum):
         SUSIE_INF (str): SuSiE-inf method implemented in `gentropy`
     """
 
-    PICS = "pics"
+    PICS = "PICS"
     SUSIE = "SuSie"
     SUSIE_INF = "SuSiE-inf"
 

diff --git a/src/gentropy/datasource/gwas_catalog/study_index.py b/src/gentropy/datasource/gwas_catalog/study_index.py
@@ -655,11 +655,7 @@ def add_no_sumstats_flag(self: StudyIndexGWASCatalog) -> StudyIndexGWASCatalog:
         """
         self.df = self.df.withColumn(
             "qualityControls",
-            StudyIndex.update_quality_flag(
-                f.col("qualityControls"),
-                ~f.col("hasSumstats"),
-                StudyQualityCheck.SUMSTATS_NOT_AVAILABLE,
-            ),
+            f.array(f.lit(StudyQualityCheck.SUMSTATS_NOT_AVAILABLE.value))
         )
         return self
 

diff --git a/src/gentropy/method/pics.py b/src/gentropy/method/pics.py
@@ -280,6 +280,15 @@ def finemap(
                         StudyLocusQualityCheck.NOT_QUALIFYING_LD_BLOCK,
                     ),
                 )
+                # Flagging all PICS loci with OUT_OF_SAMPLE_LD flag:
+                .withColumn(
+                    "qualityControls",
+                    StudyLocus.update_quality_flag(
+                        f.col("qualityControls"),
+                        f.lit(True),
+                        StudyLocusQualityCheck.OUT_OF_SAMPLE_LD,
+                    ),
+                )
                 .withColumn(
                     "finemappingMethod",
                     finemapping_method_expression,

diff --git a/tests/gentropy/dataset/test_study_locus.py b/tests/gentropy/dataset/test_study_locus.py
@@ -629,7 +629,7 @@ class TestStudyLocusValidation:
 
     STUDY_LOCUS_DATA = [
         # Won't be flagged:
-        ("1", "v1", "s1", 1.0, -8, [], "pics"),
+        ("1", "v1", "s1", 1.0, -8, [], "PICS"),
         # Already flagged, needs to be tested if the flag reamins unique:
         (
             "2",
@@ -638,7 +638,7 @@ class TestStudyLocusValidation:
             5.0,
             -4,
             [StudyLocusQualityCheck.SUBSIGNIFICANT_FLAG.value],
-            "pics",
+            "PICS",
         ),
         # To be flagged:
         ("3", "v3", "s3", 1.0, -4, [], "SuSiE-inf"),
@@ -869,18 +869,18 @@ class TestStudyLocusRedundancyFlagging:
     """Collection of tests related to flagging redundant credible sets."""
 
     STUDY_LOCUS_DATA = [
-        ("1", "v1", "s1", "pics", []),
-        ("2", "v2", "s1", "pics", [StudyLocusQualityCheck.TOP_HIT.value]),
-        ("3", "v3", "s1", "pics", []),
-        ("3", "v3", "s1", "pics", []),
-        ("1", "v1", "s1", "pics", [StudyLocusQualityCheck.TOP_HIT.value]),
-        ("1", "v1", "s2", "pics", [StudyLocusQualityCheck.TOP_HIT.value]),
-        ("1", "v1", "s2", "pics", [StudyLocusQualityCheck.TOP_HIT.value]),
+        ("1", "v1", "s1", "PICS", []),
+        ("2", "v2", "s1", "PICS", [StudyLocusQualityCheck.TOP_HIT.value]),
+        ("3", "v3", "s1", "PICS", []),
+        ("3", "v3", "s1", "PICS", []),
+        ("1", "v1", "s1", "PICS", [StudyLocusQualityCheck.TOP_HIT.value]),
+        ("1", "v1", "s2", "PICS", [StudyLocusQualityCheck.TOP_HIT.value]),
+        ("1", "v1", "s2", "PICS", [StudyLocusQualityCheck.TOP_HIT.value]),
         ("1", "v1", "s3", "SuSie", []),
-        ("1", "v1", "s3", "pics", [StudyLocusQualityCheck.TOP_HIT.value]),
-        ("1", "v1", "s4", "pics", []),
+        ("1", "v1", "s3", "PICS", [StudyLocusQualityCheck.TOP_HIT.value]),
+        ("1", "v1", "s4", "PICS", []),
         ("1", "v1", "s4", "SuSie", []),
-        ("1", "v1", "s4", "pics", [StudyLocusQualityCheck.TOP_HIT.value]),
+        ("1", "v1", "s4", "PICS", [StudyLocusQualityCheck.TOP_HIT.value]),
     ]
 
     STUDY_LOCUS_SCHEMA = t.StructType(
@@ -946,7 +946,7 @@ class TestStudyLocusSuSiERedundancyFlagging:
             "v1",
             "s1",
             "X",
-            "pics",
+            "PICS",
             1,
             3,
             [
@@ -962,7 +962,7 @@ class TestStudyLocusSuSiERedundancyFlagging:
             "v2",
             "s1",
             "X",
-            "pics",
+            "PICS",
             4,
             5,
             [
@@ -977,7 +977,7 @@ class TestStudyLocusSuSiERedundancyFlagging:
             "v3",
             "s1",
             "X",
-            "pics",
+            "PICS",
             6,
             7,
             [
@@ -1004,7 +1004,7 @@ class TestStudyLocusSuSiERedundancyFlagging:
             "v5",
             "s1",
             "X",
-            "pics",
+            "PICS",
             5,
             5,
             [
@@ -1018,7 +1018,7 @@ class TestStudyLocusSuSiERedundancyFlagging:
             "v6",
             "s2",
             "X",
-            "pics",
+            "PICS",
             3,
             5,
             [
@@ -1141,11 +1141,11 @@ class TestStudyLocusDuplicationFlagging:
 
     STUDY_LOCUS_DATA = [
         # Non-duplicated:
-        ("1", "v1", "s1", "pics"),
+        ("1", "v1", "s1", "PICS"),
         # Triplicate:
-        ("3", "v3", "s1", "pics"),
-        ("3", "v3", "s1", "pics"),
-        ("3", "v3", "s1", "pics"),
+        ("3", "v3", "s1", "PICS"),
+        ("3", "v3", "s1", "PICS"),
+        ("3", "v3", "s1", "PICS"),
     ]
 
     STUDY_LOCUS_SCHEMA = t.StructType(