Removed all commented code

src/skalo/compaction.rs

@@ -1,20 +1,14 @@
+use dashmap::DashMap;
 use hashbrown::{HashMap, HashSet};
-//use std::time::Instant;
-
 use rayon::prelude::*;
-use dashmap::DashMap;
-
-//use crate::utils::DATA_INFO;
-
 
-pub fn compact_graph(all_kmers: &mut HashMap<u128, Vec<u128>>, start_kmers: &HashSet<u128>, end_kmers: &HashSet<u128>) -> DashMap<u128, Vec<u128>> {
-
-    //let data_info = DATA_INFO.get().unwrap();
-
-    //let start = Instant::now();
-
+pub fn compact_graph(
+    all_kmers: &mut HashMap<u128, Vec<u128>>,
+    start_kmers: &HashSet<u128>,
+    end_kmers: &HashSet<u128>,
+) -> DashMap<u128, Vec<u128>> {
     let compacted: DashMap<u128, Vec<u128>> = DashMap::new();
-    
+
     // from start k-mers
     start_kmers.par_iter().for_each(|kmer| {
         if let Some(starting_kmers) = all_kmers.get(kmer) {
@@ -33,7 +27,9 @@ pub fn compact_graph(all_kmers: &mut HashMap<u128, Vec<u128>>, start_kmers: &Has
                             vec_visited.push(current_kmer);
                             visited.insert(current_kmer);
 
-                            if end_kmers.contains(&current_kmer) || start_kmers.contains(&current_kmer) {
+                            if end_kmers.contains(&current_kmer)
+                                || start_kmers.contains(&current_kmer)
+                            {
                                 walking_along_path = false;
                             }
                         } else {
@@ -68,7 +64,9 @@ pub fn compact_graph(all_kmers: &mut HashMap<u128, Vec<u128>>, start_kmers: &Has
                             vec_visited.push(current_kmer);
                             visited.insert(current_kmer);
 
-                            if end_kmers.contains(&current_kmer) || start_kmers.contains(&current_kmer) {
+                            if end_kmers.contains(&current_kmer)
+                                || start_kmers.contains(&current_kmer)
+                            {
                                 walking_along_path = false;
                             }
                         } else {
@@ -78,43 +76,37 @@ pub fn compact_graph(all_kmers: &mut HashMap<u128, Vec<u128>>, start_kmers: &Has
                         walking_along_path = false;
                     }
                 }
-                // could be "1" but for some reason I get more variant groups with k_graph
-                //if vec_visited.len() > data_info.k_graph {
                 if vec_visited.len() > 1 {
                     compacted.insert(*starting_kmer, vec_visited);
                 }
             }
         }
     });
 
-    //let duration = start.elapsed();
-    //println!("time taken: {:?}", duration);
-
-    //let start = Instant::now();
-
     // modify graph and compacted vector
-    //let mut nb_removed = 0;
     for mut item in compacted.iter_mut() {
         let (starting_kmer, vec_visited) = item.pair_mut();
         // remove edges corresponding to compacted vector
-        all_kmers.get_mut(starting_kmer).unwrap().retain(|&neighbor| neighbor != vec_visited[0]);
-        for window in vec_visited[..vec_visited.len() -1].windows(2) {
-            all_kmers.get_mut(&window[0]).unwrap().retain(|&neighbor| neighbor != window[1]);
-            //nb_removed += 1
-        }        
-
+        all_kmers
+            .get_mut(starting_kmer)
+            .unwrap()
+            .retain(|&neighbor| neighbor != vec_visited[0]);
+        for window in vec_visited[..vec_visited.len() - 1].windows(2) {
+            all_kmers
+                .get_mut(&window[0])
+                .unwrap()
+                .retain(|&neighbor| neighbor != window[1]);
+        }
+
         // add new edge in place of compacted segment
-        all_kmers.entry(*starting_kmer)
-                 .or_default()
-                 .push(vec_visited[vec_visited.len() -1]);
+        all_kmers
+            .entry(*starting_kmer)
+            .or_default()
+            .push(vec_visited[vec_visited.len() - 1]);
 
         // remove last element of compact vector
         vec_visited.pop();
     }
-
-    //let duration = start.elapsed();
-    //println!("{} edges removed", nb_removed);    
-    //println!("time taken: {:?}", duration);    
 
     compacted
 }

src/skalo/extremities.rs

@@ -1,6 +1,5 @@
-use hashbrown::{HashMap, HashSet};
-//use std::time::Instant;
 use bit_set::BitSet;
+use hashbrown::{HashMap, HashSet};
 
 use crate::skalo::utils::{rev_compl_u128, DATA_INFO};
 
@@ -15,8 +14,6 @@ pub fn identify_good_kmers(
     let mut start_kmers: HashSet<u128> = HashSet::new();
     let mut end_kmers: HashSet<u128> = HashSet::new();
 
-    //let start = Instant::now();
-
     // iterate over all_kmers
     for (kmer, next_kmers) in all_kmers.iter() {
         if next_kmers.len() > 1 {
@@ -57,9 +54,6 @@ pub fn identify_good_kmers(
 
     println!("     . {} entry nodes", start_kmers.len());
 
-    //let duration = start.elapsed();
-    //println!("time taken: {:?}", duration);
-
     (start_kmers, end_kmers)
 }
 

src/skalo/input.rs

@@ -1,6 +1,5 @@
-use hashbrown::HashMap;
-//use std::time::Instant;
 use bit_set::BitSet;
+use hashbrown::HashMap;
 
 use dashmap::DashMap;
 use rayon::prelude::*;
@@ -19,8 +18,6 @@ pub fn read_input_file() -> (usize, Vec<String>, KmerGraph, KmerSamples) {
 
     println!(" # read file {}", arguments.input_file);
 
-    //let start = Instant::now();
-
     // read the skf file and load split-kmers (ska_array), kmer length and sample names
     let ska_array = load_array::<u128>(&[arguments.input_file.to_string()], arguments.nb_threads)
         .expect("\nerror: could not read the skf file\n\n");
@@ -31,13 +28,8 @@ pub fn read_input_file() -> (usize, Vec<String>, KmerGraph, KmerSamples) {
     println!("     . {}-mers", len_kmer);
     println!("     . {} samples", sample_names.len());
 
-    //let duration = start.elapsed();
-    //println!("time taken: {:?}", duration);
-
     println!(" # build colored de Bruijn graph");
 
-    //let start = Instant::now();
-
     // build De Bruijn graph
     let degenerate_code: HashMap<u8, Vec<char>> = [
         (b'A', vec!['A']),
@@ -129,8 +121,6 @@ pub fn read_input_file() -> (usize, Vec<String>, KmerGraph, KmerSamples) {
     let all_kmers: KmerGraph = all_kmers.into_iter().collect();
     let kmer_samples: KmerSamples = kmer_samples.into_iter().collect();
 
-    //let duration = start.elapsed();
-    //println!("time taken: {:?}", duration);
     println!("     . {} nodes", all_kmers.len());
 
     (len_kmer, sample_names, all_kmers, kmer_samples)

src/skalo/positioning.rs

@@ -132,11 +132,6 @@ pub fn scan_variants(
     let mut vec_position_forward: Vec<u32> = Vec::new();
     let mut vec_position_reverse: Vec<u32> = Vec::new();
 
-    // process only the 1st and last variants
-    //let first = &vec_variants[0];
-    //let last = &vec_variants[vec_variants.len() - 1];
-
-    //for variant in [first, last] {
     for variant in vec_variants {
         let seq = variant.sequence.decode();
         let rc_seq = rev_compl(&seq);

src/skalo/process_variants.rs

@@ -1,6 +1,5 @@
-use hashbrown::{HashMap, HashSet};
-//use std::time::Instant;
 use bit_set::BitSet;
+use hashbrown::{HashMap, HashSet};
 
 use crate::skalo::output::create_fasta_and_vcf;
 use crate::skalo::positioning::{extract_genomic_kmers, scan_variants};
@@ -36,8 +35,6 @@ pub fn analyse_variant_groups(
 
     println!(" # process indels");
 
-    //let start = Instant::now();
-
     // collect entry kmers of indels
     let (final_indels, entries_indels) = process_indels(indel_groups, data_info.k_graph);
     println!("     -> {} indels", final_indels.len());
@@ -56,9 +53,6 @@ pub fn analyse_variant_groups(
         }
     }
 
-    //let duration = start.elapsed();
-    //println!("time taken: {:?}", duration);
-
     println!(" # process SNPs");
 
     // create a vector of keys sorted by the ratio of size of Vec<VariantInfo> to the length of the first sequence
@@ -76,8 +70,6 @@ pub fn analyse_variant_groups(
         .collect();
     sorted_keys.sort_by(|a, b| b.1.partial_cmp(&a.1).unwrap()); // Sort by ratio, descending
 
-    //let start = Instant::now();
-
     // start processing SNPs
     let mut entries_done: HashSet<u128> = HashSet::new();
 
@@ -102,7 +94,6 @@ pub fn analyse_variant_groups(
 
             // get SNP column and kmers
             let mut kmers_to_save: HashSet<u128> = HashSet::new();
-            //let mut found_snp_pos: HashMap<usize, Vec<char>> = HashMap::new();
             let mut found_snp_pos: HashMap<usize, Vec<char>> =
                 HashMap::with_capacity(real_snp_pos.len());
 
@@ -172,8 +163,6 @@ pub fn analyse_variant_groups(
 
                         // adjust position with SNP pos in variant group and orientation
                         for (pos, column) in found_snp_pos {
-                            //println!("{}", seq_length);
-
                             let final_position = if is_forward {
                                 position + (pos - data_info.k_graph) as u32
                             } else {
@@ -207,9 +196,6 @@ pub fn analyse_variant_groups(
         }
     }
 
-    //let duration = start.elapsed();
-    //println!("time taken: {:?}", duration);
-
     if do_postioning {
         println!(
             "     -> {} SNPs (+ {} w/o position)",

src/skalo/read_graph.rs

@@ -1,6 +1,5 @@
-use hashbrown::{HashMap, HashSet};
-//use std::time::Instant;
 use bit_set::BitSet;
+use hashbrown::{HashMap, HashSet};
 
 use indicatif::{ProgressBar, ProgressStyle};
 use rayon::prelude::*;
@@ -31,8 +30,6 @@ pub fn build_variant_groups(
 
     println!(" # explore graph");
 
-    //let start = Instant::now();
-
     let built_groups = Arc::new(Mutex::new(HashMap::<(u128, u128), Vec<VariantInfo>>::new()));
 
     let pool = ThreadPoolBuilder::new()
@@ -240,9 +237,6 @@ pub fn build_variant_groups(
         });
     });
 
-    //let duration = start.elapsed();
-    //println!("time taken: {:?}", duration);
-
     let built_groups_end = built_groups.lock().unwrap();
 
     println!("     . {} variant groups", built_groups_end.len());