Improve runtime (#44)

* expose max workers parameter to cli

* update triqler package to fix the issue with calculating PEPs which is extremely slow

* make grid search run with multithreading

* test max_workers cli argument

* fix format

* update triqler

Co-authored-by: Siegfried Gessulat <[email protected]>
Co-authored-by: Tobias <[email protected]>

mokapot/config.py

@@ -65,6 +65,14 @@ def _parser():
         ),
     )
 
+    parser.add_argument(
+        "-w",
+        "--max_workers",
+        default=1,
+        type=int,
+        help="The number of processes to use for model training.",
+    )
+
     parser.add_argument(
         "-r",
         "--file_root",

mokapot/model.py

@@ -432,7 +432,7 @@ def __init__(
         """Initialize a PercolatorModel"""
         svm_model = LinearSVC(dual=False)
         estimator = GridSearchCV(
-            svm_model, param_grid=PERC_GRID, refit=False, cv=3
+            svm_model, param_grid=PERC_GRID, refit=False, cv=3, n_jobs=-1
         )
 
         super().__init__(

mokapot/mokapot.py

@@ -92,7 +92,11 @@ def main():
 
     # Fit the models:
     psms, models = brew(
-        datasets, model=model, test_fdr=config.test_fdr, folds=config.folds
+        datasets,
+        model=model,
+        test_fdr=config.test_fdr,
+        folds=config.folds,
+        max_workers=config.max_workers,
     )
 
     if config.dest_dir is not None:

setup.cfg

@@ -25,9 +25,9 @@ install_requires =
     pandas>=1.0.3
     scikit-learn>=0.22.1
     numba>=0.48.0
-    triqler>=0.3.0
     matplotlib>=3.1.3
     lxml>=4.6.2
+    triqler>=0.6.2
 
 [options.extras_require]
 docs =

tests/system_tests/test_cli.py

@@ -69,6 +69,8 @@ def test_cli_options(tmp_path, scope_files):
         "--keep_decoys",
         "--subset_max_train",
         "50000",
+        "--max_workers",
+        "3",
     ]
 
     subprocess.run(cmd, check=True)