Move randstrobes_query to seeding

Author: marcelm

src/chainer.rs

@@ -5,9 +5,9 @@ use log::trace;
 use crate::details::NamDetails;
 use crate::hit::{Hit, HitsDetails, find_hits};
 use crate::index::StrobemerIndex;
-use crate::mapper::QueryRandstrobe;
 use crate::mcsstrategy::McsStrategy;
 use crate::nam::Nam;
+use crate::seeding::QueryRandstrobe;
 
 const N_PRECOMPUTED: usize = 1024;
 

src/hit.rs

@@ -5,8 +5,8 @@ use log::Level;
 use log::trace;
 
 use crate::index::StrobemerIndex;
-use crate::mapper::QueryRandstrobe;
 use crate::mcsstrategy::McsStrategy;
+use crate::seeding::QueryRandstrobe;
 
 #[derive(Debug)]
 pub struct Hit {

src/mapper.rs

@@ -25,7 +25,7 @@ use crate::nam::{Nam, get_nams_by_chaining, reverse_nam_if_needed};
 use crate::piecewisealigner::remove_spurious_anchors;
 use crate::read::Read;
 use crate::revcomp::reverse_complement;
-use crate::seeding::{RandstrobeIterator, SeedingParameters, SyncmerIterator};
+use crate::seeding::SeedingParameters;
 
 const MAX_PAIR_NAMS: usize = 1000;
 
@@ -76,73 +76,6 @@ impl Alignment {
     }
 }
 
-#[derive(Debug)]
-pub struct QueryRandstrobe {
-    pub hash: u64,
-    pub hash_revcomp: u64,
-    pub start: usize,
-    pub end: usize,
-}
-
-/// Generate randstrobes for a query sequence and its reverse complement.
-/// TODO move to strobes.rs?
-pub fn randstrobes_query(seq: &[u8], parameters: &SeedingParameters) -> [Vec<QueryRandstrobe>; 2] {
-    let mut randstrobes = {
-        let expected = seq.len() / (parameters.syncmer.k - parameters.syncmer.s + 1);
-        [Vec::with_capacity(expected), Vec::with_capacity(expected)]
-    };
-    if seq.len() < parameters.randstrobe.w_max {
-        return randstrobes;
-    }
-
-    // Generate syncmers for the forward sequence
-    let syncmer_iter = SyncmerIterator::new(
-        seq,
-        parameters.syncmer.k,
-        parameters.syncmer.s,
-        parameters.syncmer.t,
-    );
-    let mut syncmers: Vec<_> = syncmer_iter.collect();
-
-    // Generate randstrobes for the forward sequence
-    let randstrobe_iter =
-        RandstrobeIterator::new(syncmers.iter().cloned(), parameters.randstrobe.clone());
-
-    for randstrobe in randstrobe_iter {
-        randstrobes[0].push(QueryRandstrobe {
-            hash: randstrobe.hash,
-            hash_revcomp: randstrobe.hash_revcomp,
-            start: randstrobe.strobe1_pos,
-            end: randstrobe.strobe2_pos + parameters.syncmer.k,
-        });
-    }
-
-    // For the reverse complement, we can re-use the syncmers of the forward
-    // sequence because canonical syncmers are invariant under reverse
-    // complementing. Only the coordinates need to be adjusted.
-    syncmers.reverse();
-    for i in 0..syncmers.len() {
-        syncmers[i].position = seq.len() - syncmers[i].position - parameters.syncmer.k;
-    }
-
-    // Randstrobes cannot be re-used for the reverse complement:
-    // If in the forward direction, syncmer[i] and syncmer[j] were paired up, it
-    // is not necessarily the case that syncmer[j] is going to be paired with
-    // syncmer[i] in the reverse direction because i is fixed in the forward
-    // direction and j is fixed in the reverse direction.
-    let rc_randstrobe_iter =
-        RandstrobeIterator::new(syncmers.into_iter(), parameters.randstrobe.clone());
-    for randstrobe in rc_randstrobe_iter {
-        randstrobes[1].push(QueryRandstrobe {
-            hash: randstrobe.hash,
-            hash_revcomp: randstrobe.hash_revcomp,
-            start: randstrobe.strobe1_pos,
-            end: randstrobe.strobe2_pos + parameters.syncmer.k,
-        });
-    }
-    randstrobes
-}
-
 /// Conversion of an Alignment into a SamRecord
 #[derive(Default)]
 pub struct SamOutput {

src/nam.rs

@@ -10,9 +10,9 @@ use crate::chainer::Chainer;
 use crate::details::NamDetails;
 use crate::index::StrobemerIndex;
 use crate::io::fasta::RefSequence;
-use crate::mapper;
 use crate::mcsstrategy::McsStrategy;
 use crate::read::Read;
+use crate::seeding::randstrobes_query;
 
 /// Non-overlapping approximate match
 #[derive(Clone, Debug)]
@@ -138,7 +138,7 @@ pub fn get_nams_by_chaining(
     rng: &mut Rng,
 ) -> (NamDetails, Vec<Nam>) {
     let timer = Instant::now();
-    let query_randstrobes = mapper::randstrobes_query(sequence, &index.parameters);
+    let query_randstrobes = randstrobes_query(sequence, &index.parameters);
     let time_randstrobes = timer.elapsed().as_secs_f64();
 
     trace!(

src/seeding/mod.rs

@@ -6,3 +6,69 @@ pub mod syncmers;
 pub use parameters::{InvalidSeedingParameter, SeedingParameters};
 pub use strobes::{DEFAULT_AUX_LEN, RandstrobeIterator, RandstrobeParameters};
 pub use syncmers::{Syncmer, SyncmerIterator, SyncmerParameters};
+
+#[derive(Debug)]
+pub struct QueryRandstrobe {
+    pub hash: u64,
+    pub hash_revcomp: u64,
+    pub start: usize,
+    pub end: usize,
+}
+
+/// Generate randstrobes for a query sequence and its reverse complement.
+pub fn randstrobes_query(seq: &[u8], parameters: &SeedingParameters) -> [Vec<QueryRandstrobe>; 2] {
+    let mut randstrobes = {
+        let expected = seq.len() / (parameters.syncmer.k - parameters.syncmer.s + 1);
+        [Vec::with_capacity(expected), Vec::with_capacity(expected)]
+    };
+    if seq.len() < parameters.randstrobe.w_max {
+        return randstrobes;
+    }
+
+    // Generate syncmers for the forward sequence
+    let syncmer_iter = SyncmerIterator::new(
+        seq,
+        parameters.syncmer.k,
+        parameters.syncmer.s,
+        parameters.syncmer.t,
+    );
+    let mut syncmers: Vec<_> = syncmer_iter.collect();
+
+    // Generate randstrobes for the forward sequence
+    let randstrobe_iter =
+        RandstrobeIterator::new(syncmers.iter().cloned(), parameters.randstrobe.clone());
+
+    for randstrobe in randstrobe_iter {
+        randstrobes[0].push(QueryRandstrobe {
+            hash: randstrobe.hash,
+            hash_revcomp: randstrobe.hash_revcomp,
+            start: randstrobe.strobe1_pos,
+            end: randstrobe.strobe2_pos + parameters.syncmer.k,
+        });
+    }
+
+    // For the reverse complement, we can re-use the syncmers of the forward
+    // sequence because canonical syncmers are invariant under reverse
+    // complementing. Only the coordinates need to be adjusted.
+    syncmers.reverse();
+    for i in 0..syncmers.len() {
+        syncmers[i].position = seq.len() - syncmers[i].position - parameters.syncmer.k;
+    }
+
+    // Randstrobes cannot be re-used for the reverse complement:
+    // If in the forward direction, syncmer[i] and syncmer[j] were paired up, it
+    // is not necessarily the case that syncmer[j] is going to be paired with
+    // syncmer[i] in the reverse direction because i is fixed in the forward
+    // direction and j is fixed in the reverse direction.
+    let rc_randstrobe_iter =
+        RandstrobeIterator::new(syncmers.into_iter(), parameters.randstrobe.clone());
+    for randstrobe in rc_randstrobe_iter {
+        randstrobes[1].push(QueryRandstrobe {
+            hash: randstrobe.hash,
+            hash_revcomp: randstrobe.hash_revcomp,
+            start: randstrobe.strobe1_pos,
+            end: randstrobe.strobe2_pos + parameters.syncmer.k,
+        });
+    }
+    randstrobes
+}