update galah to include contig clustering

Author: AroneyS

Cargo.toml

@@ -32,8 +32,8 @@ version-compare = "0.2.0"
 # bird_tool_utils = "0.5.0"
 bird_tool_utils = { git = "https://github.com/wwood/bird_tool_utils" }
 # bird_tool_utils = { path = "../bird_tool_utils" }
-galah = "0.4.0"
-# galah = { git = "https://github.com/wwood/galah" }
+# galah = "0.4.0"
+galah = { git = "https://github.com/wwood/galah", rev = "a351847532c82a22743fd8e6da5643689960d46a" }
 bird_tool_utils-man = "0.4.0"
 roff = "0.2.*"
 needletail = "0.5.*"

pixi.lock

@@ -18,6 +18,7 @@ samtools = ">=1.9"
 coreutils = "*"
 minimap2 = ">=2.24,<2.29" # Restrict due to test_no_zeros_bug1_minimap2 failing with slightly different values
 bwa = ">=0.7.17"
+skani = ">=0.2.2"
 fastani = ">=1.3"
 extern = "*"
 

pixi.toml

@@ -813,6 +813,8 @@ fn dereplicate(m: &clap::ArgMatches, genome_fasta_files: &Vec<String>) -> Vec<St
     // Generate clusterer and check for dependencies
     let clusterer = galah::cluster_argument_parsing::generate_galah_clusterer(
         genome_fasta_files,
+        &None, // No additional genome files
+        false, // Not using a different mode
         m,
         &coverm::cli::COVERM_CLUSTER_COMMAND_DEFINITION,
     )
@@ -843,6 +845,7 @@ fn dereplicate(m: &clap::ArgMatches, genome_fasta_files: &Vec<String>) -> Vec<St
         cluster_outputs,
         &cluster_indices,
         &clusterer.genome_fasta_paths,
+        None, // No additional output parameter needed
     );
 
     reps

src/bin/coverm.rs

@@ -29,9 +29,13 @@ lazy_static! {
             dereplication_ani_argument: "dereplication-ani".to_string(),
             dereplication_prethreshold_ani_argument: "dereplication-prethreshold-ani".to_string(),
             dereplication_quality_formula_argument: "dereplication-quality-formula".to_string(),
-            dereplication_cluster_method_argument: "dereplication-cluster-method".to_string(),
             dereplication_precluster_method_argument: "dereplication-precluster-method".to_string(),
+            dereplication_cluster_method_argument: "dereplication-cluster-method".to_string(),
             dereplication_aligned_fraction_argument: "dereplication-aligned-fraction".to_string(),
+            dereplication_small_genomes_argument: "dereplication-small-genomes".to_string(),
+            dereplication_cluster_contigs_argument: "dereplication-cluster-contigs".to_string(),
+            dereplication_small_contigs_argument: "dereplication-small-contigs".to_string(),
+            dereplication_large_contigs_argument: "dereplication-large-contigs".to_string(),
             dereplication_fraglen_argument: "dereplication-fragment-length".to_string(),
             dereplication_output_cluster_definition_file: "dereplication-output-cluster-definition"
                 .to_string(),
@@ -1470,24 +1474,44 @@ Ben J. Woodcroft <benjwoodcroft near gmail.com>
                         ])
                         .default_value(galah::DEFAULT_QUALITY_FORMULA),
                 )
-                .arg(
-                    Arg::new("dereplication-cluster-method")
-                        .long("dereplication-cluster-method")
-                        .value_parser(galah::CLUSTER_METHODS)
-                        .default_value(galah::DEFAULT_CLUSTER_METHOD),
-                )
                 .arg(
                     Arg::new("dereplication-precluster-method")
                         .long("dereplication-precluster-method")
                         .value_parser(galah::PRECLUSTER_METHODS)
                         .default_value(galah::DEFAULT_PRECLUSTER_METHOD),
                 )
+                .arg(
+                    Arg::new("dereplication-cluster-method")
+                        .long("dereplication-cluster-method")
+                        .value_parser(galah::CLUSTER_METHODS)
+                        .default_value(galah::DEFAULT_CLUSTER_METHOD),
+                )
                 .arg(
                     Arg::new("dereplication-aligned-fraction")
                         .long("dereplication-aligned-fraction")
                         .default_value(galah::DEFAULT_ALIGNED_FRACTION)
                         .value_parser(clap::value_parser!(f32)),
                 )
+                .arg(
+                    Arg::new("dereplication-small-genomes")
+                        .long("dereplication-small-genomes")
+                        .action(clap::ArgAction::SetTrue),
+                )
+                .arg(
+                    Arg::new("dereplication-cluster-contigs")
+                        .long("dereplication-cluster-contigs")
+                        .action(clap::ArgAction::SetTrue),
+                )
+                .arg(
+                    Arg::new("dereplication-small-contigs")
+                        .long("dereplication-small-contigs")
+                        .action(clap::ArgAction::SetTrue),
+                )
+                .arg(
+                    Arg::new("dereplication-large-contigs")
+                        .long("dereplication-large-contigs")
+                        .action(clap::ArgAction::SetTrue),
+                )
                 .arg(
                     Arg::new("dereplication-fragment-length")
                         .long("dereplication-fragment-length")