From b5ac289658f523b97418c9a2314e50a0ec7231e1 Mon Sep 17 00:00:00 2001
From: Martin Larralde <martin.larralde@embl.de>
Date: Fri, 15 Dec 2023 14:10:39 +0100
Subject: [PATCH] Add trait to count the number of symbol occurences in an
arbitrary collection
---
lightmotif/src/seq.rs | 28 +++++++++++++++++++++++++---
1 file changed, 25 insertions(+), 3 deletions(-)
diff --git a/lightmotif/src/seq.rs b/lightmotif/src/seq.rs
index 984b794..8647365 100644
--- a/lightmotif/src/seq.rs
+++ b/lightmotif/src/seq.rs
@@ -18,6 +18,19 @@ use super::err::InvalidSymbol;
use super::num::StrictlyPositive;
use super::pwm::ScoringMatrix;
+// --- SymbolCount -------------------------------------------------------------
+
+/// A trait for counting the number of occurences of a symbol in a sequence.
+pub trait SymbolCount<A: Alphabet> {
+ fn count_symbol(&self, symbol: <A as Alphabet>::Symbol) -> usize;
+}
+
+impl<'a, A: Alphabet, T: IntoIterator<Item = &'a A::Symbol> + Copy> SymbolCount<A> for T {
+ fn count_symbol(&self, symbol: <A as Alphabet>::Symbol) -> usize {
+ self.into_iter().filter(|&&c| c == symbol).count()
+ }
+}
+
// --- EncodedSequence ---------------------------------------------------------
/// A biological sequence encoded with an alphabet.
@@ -53,8 +66,8 @@ impl<A: Alphabet> EncodedSequence<A> {
/// Iterate over the symbols in the sequence.
#[inline]
- pub fn iter(&self) -> impl IntoIterator<Item = &A::Symbol> {
- self.data.iter()
+ pub fn iter(&self) -> <&[A::Symbol] as IntoIterator>::IntoIter {
+ self.data.as_slice().into_iter()
}
/// Convert the encoded sequence to a striped matrix.
@@ -233,7 +246,7 @@ impl<A: Alphabet, C: StrictlyPositive> AsRef<StripedSequence<A, C>> for StripedS
}
}
-impl<A: Alphabet, C: Unsigned + NonZero> Index<usize> for StripedSequence<A, C> {
+impl<A: Alphabet, C: StrictlyPositive> Index<usize> for StripedSequence<A, C> {
type Output = <A as Alphabet>::Symbol;
#[inline]
fn index(&self, index: usize) -> &Self::Output {
@@ -244,6 +257,15 @@ impl<A: Alphabet, C: Unsigned + NonZero> Index<usize> for StripedSequence<A, C>
}
}
+impl<A: Alphabet, C: StrictlyPositive> SymbolCount<A> for &StripedSequence<A, C> {
+ fn count_symbol(&self, symbol: <A as Alphabet>::Symbol) -> usize {
+ self.data
+ .iter()
+ .map(|row| SymbolCount::<A>::count_symbol(&row, symbol))
+ .sum()
+ }
+}
+
#[cfg(test)]
mod test {
use typenum::consts::U2;
--
GitLab