bugfix regenie check

PMBio · Sep 30, 2024 · 9f37a5e · 9f37a5e
1 parent 1630243
commit 9f37a5e
Show file tree

Hide file tree

Showing 3 changed files with 52 additions and 7 deletions.
diff --git a/.github/workflows/pipeline-tests.yml b/.github/workflows/pipeline-tests.yml
@@ -67,6 +67,16 @@ jobs:
       prerun_cmd: cp ./tests/deeprvat/pretrained/deeprvat_config.yaml ./example/
       dry_run: false
       upload_pretrained_outputs: true
+
+  Reference-Output-Pretrained-Regenie:
+    uses: ./.github/workflows/run-pipeline.yml
+    with:
+      branch: main
+      pipeline_file: ./pipelines/association_testing_pretrained_regenie.snakefile
+      environment_file: ./deeprvat_env_no_gpu.yml
+      prerun_cmd: cp ./tests/deeprvat/regenie/pretrained/deeprvat_config.yaml ./example/
+      dry_run: false
+      upload_regenie_outputs: true
 
   # Association Testing Pretrained Pipeline
   Smoke-Association-Testing-Pretrained:
@@ -97,16 +107,16 @@ jobs:
       prerun_cmd: cp ./tests/deeprvat/regenie/pretrained/deeprvat_config.yaml ./example/
 
   Pipeline-Tests-Association-Testing-Pretrained-Regenie:
-    needs: [Smoke-Association-Testing-Pretrained-Regenie, Reference-Output-Pretrained]
+    needs: [Smoke-Association-Testing-Pretrained-Regenie, Reference-Output-Pretrained-Regenie]
     uses: ./.github/workflows/run-pipeline.yml
     with:
       pipeline_file: ./pipelines/association_testing_pretrained_regenie.snakefile
       environment_file: ./deeprvat_env_no_gpu.yml
       prerun_cmd: cp ./tests/deeprvat/regenie/pretrained/deeprvat_config.yaml ./example/
       dry_run: false
-      download_pretrained_outputs: true
-      run_burden_results_check: true
-      run_association_results_check: true
+      download_regenie_outputs: true
+      #run_burden_results_check: true
+      run_regenie_association_results_check: true
 
   # Association Testing Training
   Smoke-Association-Testing-Training:

diff --git a/.github/workflows/run-pipeline.yml b/.github/workflows/run-pipeline.yml
@@ -52,6 +52,10 @@ on:
         required: false
         default: false
         type: boolean
+      upload_regenie_outputs:
+        required: false
+        default: false
+        type: boolean
       download_training_outputs:
         required: false
         default: false
@@ -60,6 +64,10 @@ on:
         required: false
         default: false
         type: boolean
+      download_regenie_outputs:
+        required: false
+        default: false
+        type: boolean
       run_training_results_check:
         required: false
         default: false
@@ -72,6 +80,10 @@ on:
         required: false
         default: false
         type: boolean
+      run_regenie_association_results_check:
+        required: false
+        default: false
+        type: boolean
 
 jobs:
   Run-Pipeline:
@@ -148,6 +160,15 @@ jobs:
               ./example/**/all_results.parquet
             include-hidden-files: true #for .zarr needed
             retention-days: 1
+        - name: Upload Regenie Outputs
+          id: uploaded_regenie_outputs
+          if: inputs.upload_regenie_outputs
+          uses: actions/upload-artifact@v4
+          with:
+            name: completed_regenie_outputs
+            path: |
+              ./example/**/all_results.parquet
+            retention-days: 1
         - name: Download Previous Training Outputs
           id: downloaded_training_outputs
           if: inputs.download_training_outputs
@@ -162,6 +183,13 @@ jobs:
           with:
             name: completed_pretrained_outputs
             path: ./tests/completed_pretrained_outputs
+        - name: Download Previous Regenie Outputs
+          id: downloaded_regenie_outputs
+          if: inputs.download_regenie_outputs
+          uses: actions/download-artifact@v4
+          with:
+            name: completed_regenie_outputs
+            path: ./tests/completed_regenie_outputs
         # - name: Display structure of downloaded files
         #   if: inputs.download_outputs
         #   run: ls -R ./tests/completed_run_output
@@ -186,4 +214,11 @@ jobs:
             ./example/ ./tests/completed_pretrained_outputs/ \
             "Cholesterol" "Platelet_count"
           shell: micromamba-shell {0}
+        - name: Run REGENIE Association Results Check
+          if: inputs.run_regenie_association_results_check
+          run: |
+            python $GITHUB_WORKSPACE/tests/deeprvat/compare_reference.py compare-association \
+            ./example/ ./tests/completed_regenie_outputs/ \
+            "Cholesterol" "Platelet_count"
+          shell: micromamba-shell {0}
 
diff --git a/tests/deeprvat/compare_reference.py b/tests/deeprvat/compare_reference.py
@@ -105,11 +105,11 @@ def compare_training(
         )
 
         if max_difference > tolerance:
+            params_pairs = [(p1, p2) for p1, p2 in zip(model.parameters(), reference_model.parameters())]
+            logger.info("Model and reference model parameter pairs:\n %s", params_pairs)
             raise RuntimeError(
                 f"FAIL! Max difference between model and reference parameters (repeat {r}) "
-                f"differs by {max_difference} > {tolerance=}\n"
-                f"REFERENCE Model Params: {reference_model.parameters()}\n\n"
-                f"Current Test Model Params; {model.parameters()}"
+                f"differs by {max_difference} > {tolerance=}\n"                
             )
         else:
             logger.info(