Allow disabling cluster trimming with a CLI flag

Author: althonos

gecco/cli/commands/_common.py

@@ -589,12 +589,17 @@ def extract_clusters(
     postproc: str,
     cds: int,
     edge_distance: int,
+    trim: bool,
 ) -> List["Cluster"]:
     from ...refine import ClusterRefiner
 
     logger.info("Extracting", "predicted clusters", level=1)
     refiner = ClusterRefiner(
-        threshold=threshold, criterion=postproc, n_cds=cds, edge_distance=edge_distance
+        threshold=threshold,
+        criterion=postproc,
+        n_cds=cds,
+        edge_distance=edge_distance,
+        trim=trim,
     )
 
     with logger.progress() as progress:

gecco/cli/commands/_parser.py

@@ -306,6 +306,28 @@ def configure_group_cluster_detection(
             "of false positives on small contigs."
         ),
     )
+
+    if defaults.get("--trim", True):
+        group.add_argument(
+            "--no-trim",
+            action="store_false",
+            dest="trim",
+            help=(
+                "Disable trimming genes lacking domain annotations from gene "
+                "clusters predicted by the internal CRF model."
+            ),
+        )
+    else:
+        group.add_argument(
+            "--trim",
+            action="store_true",
+            dest="trim",
+            help=(
+                "Enable trimming genes lacking domain annotations from gene "
+                "clusters predicted by the internal CRF model."
+            ),
+        )
+
     return group
 
 

gecco/cli/commands/predict.py

@@ -106,6 +106,7 @@ def run(
         postproc=args.postproc,
         cds=args.cds,
         edge_distance=args.edge_distance,
+        trim=args.trim,
     )
     if clusters:
         logger.success("Found", len(clusters), "potential gene clusters", level=1)

gecco/cli/commands/run.py

@@ -160,6 +160,7 @@ def run(
         postproc=args.postproc,
         cds=args.cds,
         edge_distance=args.edge_distance,
+        trim=args.trim,
     )
     if clusters:
         logger.success("Found", len(clusters), "potential gene clusters", level=1)