single omics workflow updated

scripts/run_benchmark_single_omics.sh

@@ -1,12 +1,12 @@
 #!/bin/bash
 
 # RUN_ID="run_$(date +%Y-%m-%d_%H-%M-%S)"
-RUN_ID="single_omics_try1"
-resources_dir="s3://openproblems-data/resources/grn"
-publish_dir="s3://openproblems-data/resources/grn/results/${RUN_ID}"
+RUN_ID="single_omics"
+# resources_dir="s3://openproblems-data/resources_test/grn"
+# publish_dir="s3://openproblems-data/resources_test/grn/results/${RUN_ID}"
 
-# resources_dir="./resources_test/"
-# publish_dir="output/${RUN_ID}"
+resources_dir="./resources_test/"
+publish_dir="output/${RUN_ID}"
 
 reg_type=ridge
 subsample=-2

src/api/comp_metric.yaml

@@ -51,6 +51,10 @@ functionality:
       type: file
       direction: input
       default: 'resources/prior/tf_all.csv'
+    - name: --apply_tf
+      type: boolean 
+      required: false
+      default: true
 
 
 

src/methods/single_omics/tigress/config.vsh.yaml

@@ -26,4 +26,4 @@ platforms:
   - type: native
   - type: nextflow
     directives:
-      label: [midtime,midmem,midcpu]
+      label: [midtime, midmem, highcpu]

src/metrics/regression_1/config.vsh.yaml

@@ -13,6 +13,11 @@ functionality:
       direction: input
       required: false
       default: pearson
+    - name: --min_tf
+      type: integer
+      direction: input
+      description: calculate the scores for the given min tfs in addition to the default 
+      required: false
   resources:
     - type: python_script
       path: script.py

src/metrics/regression_1/main.py

@@ -177,7 +177,8 @@ def main(par):
     gene_names = perturbation_data.var.index.to_numpy()
     net = pd.read_csv(par['prediction'])
     # subset to keep only those links with source as tf
-    net = net[net.source.isin(tf_all)]
+    if par['apply_tf']:
+        net = net[net.source.isin(tf_all)]
 
     subsample = par['subsample']
     reg_type = par['reg_type']
@@ -210,13 +211,15 @@ def main(par):
     net_processed = process_net(net.copy(), gene_names, manipulate)
 
     print(f'Compute metrics for layer: {layer}', flush=True)
+    tfs_cases = [-1]
+    if par['min_tf']:
+        tfs_cases += par['min_tf']
     layer_results = {}  # Store results for this layer
-    for exclude_missing_genes in [False]:  # two settings on target gene
-        for tf_n in [-1]:  # two settings on tfs
+    for exclude_missing_genes in [False, True]:  # two settings on target gene
+        for tf_n in tfs_cases:  # two settings on tfs
             run_key = f'ex({exclude_missing_genes})_tf({tf_n})'
             print(run_key)
             net_subset = net_processed.copy()
-
             # Subset TFs 
             if tf_n == -1:
                 degrees = net_subset.abs().sum(axis=0)
@@ -234,11 +237,12 @@ def main(par):
 
     # Convert results to DataFrame
     df_results = pd.DataFrame(layer_results)
-    # if 'ex(True)_tf(140)' not in df_results.columns:
-    #     df_results['ex(True)_tf(140)'] = df_results['ex(True)_tf(-1)']
-    # if 'ex(False)_tf(140)' not in df_results.columns:
-    #     df_results['ex(False)_tf(140)'] = df_results['ex(False)_tf(-1)']
-
+    if par['min_tf']:
+        if 'ex(True)_tf(140)' not in df_results.columns:
+            df_results['ex(True)_tf(140)'] = df_results['ex(True)_tf(-1)']
+        if 'ex(False)_tf(140)' not in df_results.columns:
+            df_results['ex(False)_tf(140)'] = df_results['ex(False)_tf(-1)']
+
     df_results['Mean'] = df_results.mean(axis=1)
 
     return df_results

src/metrics/regression_2/main.py

@@ -277,8 +277,9 @@ def main(par: Dict[str, Any]) -> pd.DataFrame:
     n_features_theta_max = np.asarray([data[gene_name]['1'] for gene_name in gene_names], dtype=int)
 
     # Load list of putative TFs
-    df = pd.read_csv(par['tf_all'], header=None, names=['gene_name'])
-    tf_names = set(list(df['gene_name'].to_numpy()))
+    tf_names = np.loadtxt(par['tf_all'], dtype=str)
+    if par['apply_tf']==False:
+        tf_names = gene_names
 
     # Evaluate GRN
     print(f'Compute metrics for layer: {layer}', flush=True)

src/process_data/test_data/config.vsh.yaml

@@ -9,33 +9,33 @@ functionality:
   arguments:
     - name: --multiomics_rna
       type: file 
-      required: true
+      required: false
       direction: input
       default: resources/grn-benchmark/multiomics_rna.h5ad
     - name: --multiomics_rna_test
       type: file 
-      required: true
+      required: false
       direction: output
       default: resources_test/grn-benchmark/multiomics_rna.h5ad
 
     - name: --multiomics_atac
       type: file 
-      required: true
+      required: false
       direction: input
       default: resources/grn-benchmark/multiomics_atac.h5ad
     - name: --multiomics_atac_test
       type: file 
-      required: true
+      required: false
       direction: input
       default: resources_test/grn-benchmark/multiomics_atac.h5ad
     - name: --perturbation_data
       type: file 
-      required: true
+      required: false
       direction: input
       default: resources/grn-benchmark/perturbation_data.h5ad
     - name: --perturbation_data_test
       type: file 
-      required: true
+      required: false
       direction: output
       default: resources_test/grn-benchmark/perturbation_data.h5ad
   resources:

src/process_data/test_data/script.py

@@ -68,4 +68,4 @@
 
 # shorten perturbation
 adata_bulk = ad.read_h5ad(par['perturbation_data'])
-adata_bulk[:200, adata_bulk.var_names.isin(adata_rna_s.var_names)].write(par['perturbation_data_test'])
+adata_bulk[:600, adata_bulk.var_names.isin(adata_rna_s.var_names)].write(par['perturbation_data_test'])

src/workflows/run_benchmark_single_omics/main.nf

@@ -25,14 +25,17 @@ workflow run_wf {
 
   methods = [
     portia,
-    pidc,
+    ennet,
+    grnboost2,
+    scsgl,    
     ppcor,
     tigress
   ]
 
   // construct list of metrics
   metrics = [
-    regression_1
+    regression_1,
+    regression_2
   ]
 
   /****************************