update wf

src/task/workflows/run_benchmark/config.vsh.yaml

@@ -52,12 +52,22 @@ functionality:
           required: true
           direction: output
           default: task_info.yaml
-    - name: Methods
+    - name: Arguments
       arguments:
         - name: "--method_ids"
           type: string
           multiple: true
           description: A list of method ids to run. If not specified, all methods will be run.
+        - name: --bootstrap_num_replicates
+          type: integer
+          required: true
+          example: 10
+          description: Number of bootstraps to generate.
+        - name: --bootstrap_sample_fraction
+          type: double
+          required: true
+          example: 0.95
+          description: Fraction of the dataset to include in each bootstrap.
   resources:
     - type: nextflow_script
       path: main.nf
@@ -82,6 +92,7 @@ functionality:
     - name: metrics/mean_rowwise_error
     - name: metrics/mean_cosine_sim
     - name: metrics/mean_correlation
+    - name: process_dataset/bootstrap
   repositories:
     - name: openproblemsv2
       type: github

src/task/workflows/run_benchmark/main.nf

@@ -64,6 +64,37 @@ workflow run_wf {
    ***************************/
   score_ch = dataset_ch
 
+    // Bootstrap the dataset (if necessary)
+    | bootstrap.run(
+      runIf: { id, state -> state.bootstrap_fraction },
+
+      fromState: [
+        input_parquet: "de_train",
+        input_h5ad: "de_train_h5ad",
+        bootstrap_num_replicates: "bootstrap_num_replicates",
+        bootstrap_sample_fraction: "bootstrap_sample_fraction"
+      ],
+
+      toState: [
+        de_train: "output_parquet",
+        de_train_h5ad: "output_h5ad"
+      ]
+    )
+
+    // flatten bootstraps (if necessary)
+    | flatMap { id, state -> 
+      if (!state.bootstrap_fraction) {
+        return [[id, state]]
+      }
+
+      [state.de_train, state.de_train_h5ad]
+        .transpose()
+        .withIndex()
+        .collect{ el, idx ->
+          [id + "_bootstrap" + idx, state + [de_train: el[0], de_train_h5ad: el[1]]]
+        }
+    }
+
     // run all methods
     | runEach(
       components: methods,