Add in evaluation config section with correction_method and alpha par…

…ameters
PMBio · May 28, 2024 · f56b90a · f56b90a
1 parent 43fc84f
commit f56b90a
Show file tree

Hide file tree

Showing 6 changed files with 25 additions and 31 deletions.
diff --git a/deeprvat/deeprvat/config.py b/deeprvat/deeprvat/config.py
@@ -68,6 +68,7 @@ def create_main_config(
         "training",
         "n_repeats",
         "y_transformation",
+        "evaluation",
         "cv_exp",
         "cv_path",
         "n_folds",
@@ -96,7 +97,7 @@ def create_main_config(
         if input_config["use_pretrained_models"]:
             no_pretrain = False
             logger.info("Pretrained Model setup specified.")
-            to_remove = {"training", "phenotypes_for_training"}
+            to_remove = {"training", "phenotypes_for_training", "seed_gene_results"}
             expected_input_keys = [
                 item for item in expected_input_keys if item not in to_remove
             ]
@@ -200,11 +201,10 @@ def create_main_config(
         full_config["assocation_testing_data"]["dataset_config"]["rare_embedding"][
             "config"
         ]["thresholds"][k] = f"{k} {v}"
-    # Baseline results
-    full_config["baseline_results"]["options"] = input_config["seed_gene_results"][
-        "options"
-    ]
-    full_config["alpha"] = input_config["seed_gene_results"]["alpha"]
+    # Results evaluation parameters; alpha parameter for significance threshold
+    full_config["evaluation"] = {}
+    full_config["evaluation"]["correction_method"] = input_config["evaluation"]["correction_method"]
+    full_config["evaluation"]["alpha"] = input_config["evaluation"]["alpha"]
     # DeepRVAT model
     full_config["n_repeats"] = input_config["n_repeats"]
 
@@ -218,6 +218,11 @@ def create_main_config(
         full_config["training"]["early_stopping"] = input_config["training"]["early_stopping"]
         # Training Phenotypes
         full_config["training"]["phenotypes"] = input_config["phenotypes_for_training"]
+        # Baseline results
+        full_config["baseline_results"]["options"] = input_config["seed_gene_results"][
+            "options"
+        ]
+        full_config["baseline_results"]["alpha_seed_genes"] = input_config["seed_gene_results"]["alpha_seed_genes"]
     else:
         full_config["model"] = input_config["model"]
 
@@ -319,7 +324,7 @@ def update_config(
             else:
                 logger.info("Not performing EAC filtering of baseline results")
             logger.info(f"  Correcting p-values using {correction_method} method")
-            alpha = config.get("alpha_seed_genes", config.get("alpha"))
+            alpha = config["baseline_results"].get("alpha_seed_genes", config.get("alpha"))
             baseline_df = pval_correction(
                 baseline_df, alpha, correction_type=correction_method
             )

diff --git a/deeprvat/deeprvat/evaluate.py b/deeprvat/deeprvat/evaluate.py
@@ -330,7 +330,6 @@ def evaluate_(
 @click.option("--debug", is_flag=True)
 @click.option("--phenotype", type=str)
 @click.option("--use-baseline-results", is_flag=True)
-@click.option("--correction-method", type=str, default="Bonferroni")
 @click.option(
     "--combine-pval", type=str, default="Bonferroni"
 )  # Bonferroni min pval per gene for multiple baseline tests
@@ -341,7 +340,6 @@ def evaluate(
     debug: bool,
     phenotype: Optional[str],
     use_baseline_results: bool,
-    correction_method: str,
     association_files: Tuple[str],
     config_file: str,
     out_dir: str,
@@ -361,7 +359,8 @@ def evaluate(
     )
     associations["phenotype"] = pheno
 
-    alpha = config["alpha"]
+    alpha = config["evaluation"]["alpha"]
+    correction_method = config["evaluation"]["correction_method"]
 
     if use_baseline_results:
         logger.info("Reading baseline results")

diff --git a/example/config/deeprvat_input_config.yaml b/example/config/deeprvat_input_config.yaml
@@ -123,7 +123,7 @@ seed_gene_results: #baseline_results
           base: baseline_results
           type: missense/skat
 
-  alpha: 0.05
+  alpha_seed_genes: 0.05
 
 #DeepRVAT training settings
 training:
@@ -144,6 +144,11 @@ training:
 n_repeats: 6
 y_transformation: quantile_transform
 
+# Results evaluation settings
+evaluation:
+    correction_method: Bonferroni
+    alpha: 0.05
+
 #Additional settings if using the CV pipeline
 cv_exp: False
 #cv_path: sample_files

diff --git a/example/config/deeprvat_input_pretrained_models_config.yaml b/example/config/deeprvat_input_pretrained_models_config.yaml
@@ -61,28 +61,15 @@ association_testing_data_thresholds:
   MAF: "< 1e-3"
   CADD_PHRED: "> 5"
 
-#Seed Gene Baseline data settings
-seed_gene_results: #baseline_results
-  options:
-      -
-          base: baseline_results
-          type: plof/burden
-      -
-          base: baseline_results
-          type: missense/burden
-      -
-          base: baseline_results
-          type: plof/skat
-      -
-          base: baseline_results
-          type: missense/skat
-
-  alpha: 0.05
-
 #DeepRVAT model settings
 n_repeats: 30
 y_transformation: quantile_transform
 
+# Results evaluation settings
+evaluation:
+    correction_method: Bonferroni
+    alpha: 0.05
+
 #Additional settings if using the CV pipeline
 cv_exp: False
 #cv_path: sample_files

diff --git a/pipelines/association_testing/regress_eval.snakefile b/pipelines/association_testing/regress_eval.snakefile
@@ -19,7 +19,6 @@ rule evaluate:
         'deeprvat_evaluate '
         + debug +
         '{params.use_baseline_results} '
-        '--correction-method Bonferroni '
         '--phenotype {wildcards.phenotype} '
         '{input.associations} '
         '{input.config} '

diff --git a/pipelines/association_testing/regress_eval_regenie.snakefile b/pipelines/association_testing/regress_eval_regenie.snakefile
@@ -43,7 +43,6 @@ rule evaluate:
         'deeprvat_evaluate '
         + debug +
         '{params.use_baseline_results} '
-        '--correction-method Bonferroni '
         '--phenotype {wildcards.phenotype} '
         '{input.associations} '
         '{input.config} '