feat: add subsampling of phenotype values

tatonetti-lab · Aug 28, 2024 · 6ba339a · 6ba339a
1 parent 1745f23
commit 6ba339a
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 1 deletion.
diff --git a/src/webgwas_backend/main.py b/src/webgwas_backend/main.py
@@ -125,8 +125,9 @@ def validate_phenotype(
 def get_phenotype_summary(
     cohort: Annotated[Cohort, Depends(validate_cohort)],
     phenotype_definition: Annotated[ValidPhenotype, Depends(validate_phenotype)],
+    n_samples: int = 1000,
 ) -> PhenotypeSummary:
-    return get_phenotype_summary_impl(cohort, phenotype_definition)
+    return get_phenotype_summary_impl(cohort, phenotype_definition).subsample(n_samples)
 
 
 @app.post(

diff --git a/src/webgwas_backend/models.py b/src/webgwas_backend/models.py
@@ -120,3 +120,19 @@ class PhenotypeSummary(SQLModel):
         ),
     )
     rsquared: float = Field(..., description="R-squared of the phenotype definition")
+
+    def subsample(self, n_samples: int) -> "PhenotypeSummary":
+        phenotypes = [
+            ApproximatePhenotypeValues(
+                t=p.t,
+                a=p.a,
+                n=p.n,
+            )
+            for p in self.phenotypes[:n_samples]
+        ]
+        return PhenotypeSummary(
+            phenotype_definition=self.phenotype_definition,
+            cohort_name=self.cohort_name,
+            phenotypes=phenotypes,
+            rsquared=self.rsquared,
+        )