adjust default threshold-bp

sourmash-bio · Aug 24, 2023 · 8203323 · 8203323
1 parent e90fb95
commit 8203323
Showing 1 changed file with 20 additions and 10 deletions.
diff --git a/src/python/pyo3_branchwater/__init__.py b/src/python/pyo3_branchwater/__init__.py
@@ -16,15 +16,17 @@ def __init__(self, p):
                        help="a text file containing paths to .sig/.sig.gz files")
         p.add_argument('against_paths',
                        help="a text file containing paths to .sig/.sig.gz files")
-        p.add_argument('-o', '--output', required=True)
+        p.add_argument('-o', '--output', required=True,
+                       help='CSV output file for matches')
         p.add_argument('-t', '--threshold', default=0.01, type=float,
-                       help="containment threshold for matches")
+                       help='containment threshold for reporting matches')
         p.add_argument('-k', '--ksize', default=31, type=int,
-                       help="k-mer size for which to load sketches & do search")
+                       help='k-mer size at which to select sketches')
         p.add_argument('-s', '--scaled', default=1000, type=int,
-                       help="scaled value for which to load sketches & do search")
+                       help='scaled factor at which to do comparisons')
 
     def main(self, args):
+        notify(f"ksize: {args.ksize} / scaled: {args.scaled} / threshold: {args.threshold}")
         num_threads = pyo3_branchwater.get_num_threads()
         notify(f"searching all sketches in '{args.query_paths}' against '{args.against_paths}' using {num_threads} threads")
         super().main(args)
@@ -51,11 +53,15 @@ def __init__(self, p):
                        help="save gather output (minimum metagenome cover) to this file")
         p.add_argument('--output-prefetch',
                        help="save prefetch output (all overlaps) to this file")
-        p.add_argument('-t', '--threshold-bp', default=100000, type=float)
-        p.add_argument('-k', '--ksize', default=31, type=int)
-        p.add_argument('-s', '--scaled', default=1000, type=int)
+        p.add_argument('-t', '--threshold-bp', default=50000, type=float,
+                       help='threshold in estimated base pairs, for reporting matches (default: 50kb)')
+        p.add_argument('-k', '--ksize', default=31, type=int,
+                       help='k-mer size at which to do comparisons (default: 31)')
+        p.add_argument('-s', '--scaled', default=1000, type=int,
+                       help='scaled factor at which to do comparisons (default: 1000)')
 
     def main(self, args):
+        notify(f"ksize: {args.ksize} / scaled: {args.scaled} / threshold bp: {args.threshold_bp}")
         num_threads = pyo3_branchwater.get_num_threads()
         notify(f"gathering all sketches in '{args.query_sig}' against '{args.against_paths}' using {num_threads} threads")
         super().main(args)
@@ -81,11 +87,15 @@ def __init__(self, p):
         super().__init__(p)
         p.add_argument('query_paths', help="a text file containing paths to .sig/.sig.gz files to query")
         p.add_argument('against_paths', help="a text file containing paths to .sig/.sig.gz files to search against")
-        p.add_argument('-t', '--threshold-bp', default=100000, type=float)
-        p.add_argument('-k', '--ksize', default=31, type=int)
-        p.add_argument('-s', '--scaled', default=1000, type=int)
+        p.add_argument('-t', '--threshold-bp', default=50000, type=float,
+                       help='threshold in estimated base pairs, for reporting matches (default: 50kb)')
+        p.add_argument('-k', '--ksize', default=31, type=int,
+                       help='k-mer size at which to do comparisons (default: 31)')
+        p.add_argument('-s', '--scaled', default=1000, type=int,
+                       help='scaled factor at which to do comparisons (default: 1000)')
 
     def main(self, args):
+        notify(f"ksize: {args.ksize} / scaled: {args.scaled} / threshold bp: {args.threshold_bp}")
         num_threads = pyo3_branchwater.get_num_threads()
         notify(f"gathering all sketches in '{args.query_paths}' against '{args.against_paths}' using {num_threads} threads")
         super().main(args)