Merge pull request #297 from jonasscheid/add_ms2rescore

Expose deeplc calibration set size param
nf-core · Jan 4, 2024 · 992ac4f · 992ac4f
2 parents 1fc7268 + 90409ac
commit 992ac4f
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 2 deletions.
diff --git a/bin/ms2rescore_cli.py b/bin/ms2rescore_cli.py
@@ -24,7 +24,7 @@ def parse_cli_arguments_to_config(**kwargs):
 
     for key, value in kwargs.items():
         # Skip these arguments since they need to set in a nested dict of feature_generators
-        if key in ["ms2pip_model", "ms2_tolerance", "rng"]:
+        if key in ["ms2pip_model", "ms2_tolerance", "rng", "calibration_set_size"]:
             continue
 
         elif key == "feature_generators":
@@ -39,7 +39,10 @@ def parse_cli_arguments_to_config(**kwargs):
                     "ms2_tolerance": kwargs["ms2_tolerance"],
                 }
             if "deeplc" in feature_generators:
-                config["ms2rescore"]["feature_generators"]["deeplc"] = {"deeplc_retrain": False}
+                config["ms2rescore"]["feature_generators"]["deeplc"] = {
+                    "deeplc_retrain": False,
+                    "calibration_set_size": kwargs["calibration_set_size"],
+                }
             if "maxquant" in feature_generators:
                 config["ms2rescore"]["feature_generators"]["maxquant"] = {}
             if "ionmob" in feature_generators:
@@ -144,6 +147,12 @@ def filter_out_artifact_psms(
 @click.option(
     "-ms2tol", "--ms2_tolerance", help="Fragment mass tolerance [Da](default: `0.02`)", type=float, default=0.02
 )
+@click.option(
+    "-cs",
+    "--calibration_set_size",
+    help="Percentage of number of calibration set for DeepLC (default: `0.15`)",
+    default=0.15,
+)
 @click.option("-re", "--rescoring_engine", help="Either mokapot or percolator (default: `mokapot`)", default="mokapot")
 @click.option(
     "-rng", "--rng", help="Seed for mokapot's random number generator (default: `4711`)", type=int, default=4711

diff --git a/nextflow.config b/nextflow.config
@@ -73,6 +73,7 @@ params {
     rescoring_engine                = 'percolator'
     feature_generators              = 'deeplc,ms2pip'
     ms2pip_model                    = 'Immuno-HCD'
+    deeplc_calibration_set_size     = 0.15
 
     // MultiQC options
     skip_multiqc                    = false

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -296,6 +296,12 @@
                     "default": "Immuno-HCD",
                     "description": "Specify the ms2pip model that should be used for rescoring. Checkout the ms2pip documentation for available models."
                 },
+                "deeplc_calibration_set_size": {
+                    "type": "number",
+                    "fa_icon": "fas fa-file-code",
+                    "default": 0.15,
+                    "description": "Specify the number or percentage of PSMs that should be used for calibration of the deeplc model."
+                },
                 "refine_fdr_on_predicted_subset": {
                     "type": "boolean",
                     "fa_icon": "fas fa-arrows-repeat",