feat: added method to recover an unaligned dataset

crates/abd-clam/src/msa/quality/mod.rs

@@ -6,6 +6,32 @@ use rayon::prelude::*;
 
 use crate::{utils, Dataset, FlatVec};
 
+impl<U: Number, M> FlatVec<String, U, M> {
+    /// Remove all gaps from all sequences in the MSA.
+    #[must_use]
+    pub fn remove_gaps(mut self) -> Self {
+        self.instances = self
+            .instances
+            .into_iter()
+            .map(|s| s.chars().filter(|&c| c != '-').collect::<String>())
+            .collect();
+        self
+    }
+}
+
+impl<U: Number, M: Send + Sync> FlatVec<String, U, M> {
+    /// Parallel version of `remove_gaps`.
+    #[must_use]
+    pub fn par_remove_gaps(mut self) -> Self {
+        self.instances = self
+            .instances
+            .into_par_iter()
+            .map(|s| s.chars().filter(|&c| c != '-').collect::<String>())
+            .collect();
+        self
+    }
+}
+
 // TODO: Consider adding a new trait for MSA datasets. Then move these methods
 // to that trait.
 

crates/results/msa/src/main.rs

@@ -35,6 +35,10 @@ struct Args {
     #[arg(short('i'), long)]
     inp_path: PathBuf,
 
+    /// Whether the original fasta file was pre-aligned by the provider.
+    #[arg(short('p'), long)]
+    pre_aligned: bool,
+
     /// The number of samples to use for the dataset.
     #[arg(short('n'), long)]
     num_samples: Option<usize>,
@@ -103,7 +107,8 @@ fn main() -> Result<(), String> {
     ftlog::info!("Input file: {:?}", fasta_file.raw_path());
     ftlog::info!("Output directory: {:?}", fasta_file.out_dir());
 
-    let data = fasta_file.read(args.num_samples)?;
+    let data = fasta_file.read::<i32>(args.num_samples)?;
+    let data = if args.pre_aligned { data } else { data.par_remove_gaps() };
     ftlog::info!(
         "Finished reading original dataset: length range = {:?}",
         data.dimensionality_hint()