Change min freq in distance default to 0.9

bacpop · Sep 25, 2024 · 77532e2 · 77532e2
1 parent f26e7d8
commit 77532e2
Show file tree

Hide file tree

Showing 5 changed files with 21 additions and 17 deletions.
diff --git a/src/cli.rs b/src/cli.rs
@@ -11,6 +11,8 @@ pub const DEFAULT_KMER: usize = 17;
 pub const DEFAULT_PROPORTION_READS: Option<f64> = None;
 /// Default single strand (which is equivalent to !rc)
 pub const DEFAULT_STRAND: bool = false;
+/// Default minimum frequency filter threshold
+pub const DEFAULT_MINFREQ: f64 = 0.9;
 /// Default behaviour when min-freq counting ambig sites
 pub const DEFAULT_AMBIGMISSING: bool = false;
 /// Default repeat masking behaviour
@@ -191,7 +193,7 @@ pub enum Commands {
         output: Option<String>,
 
         /// Minimum fraction of samples a k-mer has to appear in
-        #[arg(short, long, value_parser = zero_to_one, default_value_t = 0.9)]
+        #[arg(short, long, value_parser = zero_to_one, default_value_t = DEFAULT_MINFREQ)]
         min_freq: f64,
 
         /// With min_freq, only count non-ambiguous sites
@@ -312,7 +314,7 @@ pub enum Commands {
         reverse: bool,
 
         /// Minimum fraction of samples a k-mer has to appear in
-        #[arg(short, long, value_parser = zero_to_one, default_value_t = 0.0)]
+        #[arg(short, long, value_parser = zero_to_one, default_value_t = DEFAULT_MINFREQ)]
         min_freq: f64,
 
         /// With min_freq, only count non-ambiguous sites

diff --git a/src/lib.rs b/src/lib.rs
@@ -209,8 +209,8 @@
 //! ska distance -o distances.txt seqs.skf
 //! ```
 //!
-//! Consider ambiguous bases by adding `--allow-ambiguous` flag, and `--min-freq` to
-//! ignore k-mers only found in some samples. Note that ambiguous bases may overestimate
+//! Consider ambiguous bases by adding `--allow-ambiguous` flag, and change `--min-freq` to
+//! ignore more/less k-mers only found in some samples (default = 0.9). Note that ambiguous bases may overestimate
 //! distances due to repeat k-mers. For finer control over filtering, first run `ska weed`
 //! on the input .skf.
 //!

diff --git a/tests/distance.rs b/tests/distance.rs
@@ -97,22 +97,24 @@ fn multisample_dists() {
         .assert()
         .success();
 
+    // Test with filters off
     Command::new(cargo_bin("ska"))
         .current_dir(sandbox.get_wd())
         .arg("distance")
         .arg("multidist.skf")
         .arg("-v")
+        .args(&["--min-freq", "0"])
+        .arg("--allow-ambiguous")
         .args(&["--threads", "2"])
         .assert()
         .stdout_eq_path(sandbox.file_string("multidist.stdout", TestDir::Correct));
 
+    // Test with default filters
     Command::new(cargo_bin("ska"))
         .current_dir(sandbox.get_wd())
         .arg("distance")
         .arg("multidist.skf")
         .arg("-v")
-        .arg("--allow-ambiguous")
-        .args(&["--min-freq", "0.5"])
         .assert()
         .stdout_eq_path(sandbox.file_string("multidist.filter.stdout", TestDir::Correct));
 }
diff --git a/tests/test_results_correct/multidist.filter.stdout b/tests/test_results_correct/multidist.filter.stdout
@@ -1,16 +1,16 @@
 Sample1	Sample2	Distance	Mismatches
-N_test_1	N_test_2	2.00	0.22222
+N_test_1	N_test_2	2.00	0.51724
 N_test_1	ambig_test_1	0.00	1.00000
 N_test_1	ambig_test_2	0.00	1.00000
-N_test_1	test_1	0.50	0.08333
-N_test_1	test_2	1.00	0.05556
+N_test_1	test_1	1.00	0.25455
+N_test_1	test_2	1.00	0.42623
 N_test_2	ambig_test_1	0.00	1.00000
 N_test_2	ambig_test_2	0.00	1.00000
-N_test_2	test_1	1.50	0.19444
-N_test_2	test_2	1.00	0.16667
-ambig_test_1	ambig_test_2	0.00	0.00000
+N_test_2	test_1	2.00	0.56716
+N_test_2	test_2	1.00	0.28571
+ambig_test_1	ambig_test_2	1.00	0.44444
 ambig_test_1	test_1	0.00	1.00000
 ambig_test_1	test_2	0.00	1.00000
 ambig_test_2	test_1	0.00	1.00000
 ambig_test_2	test_2	0.00	1.00000
-test_1	test_2	1.50	0.02778
+test_1	test_2	3.00	0.44118
diff --git a/tests/test_results_correct/multidist.stdout b/tests/test_results_correct/multidist.stdout
@@ -2,15 +2,15 @@ Sample1	Sample2	Distance	Mismatches
 N_test_1	N_test_2	2.00	0.51724
 N_test_1	ambig_test_1	0.00	1.00000
 N_test_1	ambig_test_2	0.00	1.00000
-N_test_1	test_1	1.00	0.25455
+N_test_1	test_1	0.50	0.25455
 N_test_1	test_2	1.00	0.42623
 N_test_2	ambig_test_1	0.00	1.00000
 N_test_2	ambig_test_2	0.00	1.00000
-N_test_2	test_1	2.00	0.56716
+N_test_2	test_1	1.50	0.56716
 N_test_2	test_2	1.00	0.28571
-ambig_test_1	ambig_test_2	1.00	0.44444
+ambig_test_1	ambig_test_2	0.50	0.44444
 ambig_test_1	test_1	0.00	1.00000
 ambig_test_1	test_2	0.00	1.00000
 ambig_test_2	test_1	0.00	1.00000
 ambig_test_2	test_2	0.00	1.00000
-test_1	test_2	3.00	0.44118
+test_1	test_2	2.50	0.44118