rearrange extra codec modules, Maskable trait

bio-seq/src/codec/degenerate/dna.rs

@@ -30,6 +30,13 @@ impl ReverseComplement for Seq<Dna> {
 
 #[cfg(test)]
 mod tests {
-    use crate::codec::masked;
+    use crate::codec::degenerate;
     use crate::prelude::*;
+
+    #[test]
+    fn test_1bit() {
+        let seq = Seq::<degenerate::Dna>::from_str("SSSWWWSW").unwrap();
+        let seq_rc: Seq<degenerate::Dna> = seq.revcomp();
+        assert_eq!("WSWWWSSS", String::from(seq_rc));
+    }
 }

bio-seq/src/codec/degenerate/mod.rs

@@ -1,5 +1,4 @@
-//! Experimental encodings with degenerate representations
-//! Includes `N`, `n`, `.`, `-`
+//! Experimental encodings for degenerate representations (eg 1-bit)
 
 pub mod dna;
 
@@ -9,8 +8,6 @@ pub mod dna;
 pub use dna::Dna;
 //pub use iupac::Iupac;
 
-#[cfg(test)]
-mod tests {
-    use crate::codec::masked;
-    use crate::prelude::*;
-}
+//#[cfg(test)]
+//mod tests {
+//}

bio-seq/src/codec/masked/dna.rs

@@ -1,6 +1,10 @@
-use crate::codec::Codec;
+use crate::codec::masked::Maskable;
+use crate::codec::{Codec, Complement};
 use crate::seq::{ReverseComplement, Seq};
 
+/// **Experimental** 4-bit nucleotide encoding with fast reverse complement and toggled mask operation
+///
+/// Note that masking/unmasking are not idempotent
 #[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Codec)]
 #[bits(4)]
 #[repr(u8)]
@@ -38,26 +42,32 @@ pub enum Dna {
     Unknown2 = 0b1001,
 }
 
-/*
 impl Complement for Dna {
     /// This representation can be complemented by reversing the bit pattern
     fn comp(&self) -> Self {
-        // reverse the bits
-        Dna::unsafe_from_bits(b)
+        todo!()
     }
 }
-*/
 
-impl Dna {
-    /// Flipping the bit pattern masks/unmasks this representation
-    pub fn mask(&self) -> Self {
+impl Maskable for Dna {
+    /// Inverting the bit pattern masks/unmasks this representation
+    fn mask(&self) -> Self {
+        let b = *self as u8 ^ 0b1111;
+        Dna::unsafe_from_bits(b)
+    }
+
+    fn unmask(&self) -> Self {
         let b = *self as u8 ^ 0b1111;
         Dna::unsafe_from_bits(b)
     }
 }
 
-impl Seq<Dna> {
-    pub fn mask(&self) -> Self {
+impl Maskable for Seq<Dna> {
+    fn mask(&self) -> Self {
+        Self::from(!self.bv.clone())
+    }
+
+    fn unmask(&self) -> Self {
         Self::from(!self.bv.clone())
     }
 }
@@ -75,6 +85,7 @@ impl ReverseComplement for Seq<Dna> {
 #[cfg(test)]
 mod tests {
     use crate::codec::masked;
+    use crate::codec::masked::Maskable;
     use crate::prelude::*;
 
     #[test]

bio-seq/src/codec/masked/iupac.rs

@@ -90,10 +90,21 @@ impl Maskable for Iupac {
     }
 }
 
+impl Maskable for Seq<Iupac> {
+    fn mask(&self) -> Self {
+        // set the third bit of every chunk of 5 bits
+        todo!()
+    }
+
+    fn unmask(&self) -> Self {
+        todo!()
+    }
+}
+
 impl ReverseComplement for Seq<Iupac> {
     type Output = Self;
 
-    /// A reverse complementing just requires reversing the bit sequence
+    /// Reverse complementing just requires reversing the bit sequence
     fn revcomp(&self) -> Self {
         let mut bv = self.bv.clone();
         bv.reverse();
@@ -104,5 +115,15 @@ impl ReverseComplement for Seq<Iupac> {
 #[cfg(test)]
 mod tests {
     use crate::codec::masked;
+    use crate::codec::masked::Maskable;
     use crate::prelude::*;
+
+    #[ignore]
+    #[test]
+    fn mask_iupac_seq() {
+        let seq = Seq::<masked::Iupac>::try_from("A.TCGCgtcataN--A").unwrap();
+
+        assert_ne!(seq.mask().to_string(), "a.tcgcGTGATAN--a".to_string());
+        assert_eq!(seq.mask().to_string(), "a.tcgcGTCATAn--a".to_string());
+    }
 }

bio-seq/src/codec/masked/mod.rs

@@ -11,8 +11,9 @@ pub trait Maskable {
     fn unmask(&self) -> Self;
 }
 
+/*
 #[cfg(test)]
 mod tests {
-    use crate::codec::masked;
     use crate::prelude::*;
 }
+*/