feat: move progress bar into bamchunk iterator, unify progress bar.

bio-io/Cargo.toml

@@ -13,8 +13,8 @@ version = "0.2.0"
 anyhow = "1.0.58"
 colored = "2.0.0"
 gzp = "0.11.3"
+rayon = "1.5"
 indicatif = {version = "0.17.0", features = ["rayon"]}
-#env_logger = "0.9.0"
 itertools = "0.10.5"
 lazy_static = "1.4.0"
 log = "0.4"

bio-io/src/lib.rs

@@ -6,6 +6,7 @@ use indicatif::{style, ProgressBar};
 use itertools::Itertools;
 use lazy_static::lazy_static;
 use niffler::get_reader;
+use rayon::current_num_threads;
 use regex::Regex;
 use rust_htslib::bam;
 use rust_htslib::bam::record::Aux;
@@ -182,7 +183,8 @@ pub struct BamChunk<'a> {
 }
 
 impl<'a> BamChunk<'a> {
-    pub fn new(bam: bam::Records<'a, bam::Reader>, chunk_size: usize) -> Self {
+    pub fn new(bam: bam::Records<'a, bam::Reader>, chunk_size: Option<usize>) -> Self {
+        let chunk_size = chunk_size.unwrap_or_else(|| current_num_threads() * 100);
         let bar = no_length_progress_bar();
         Self {
             bam,

src/extract.rs

@@ -1,6 +1,6 @@
 use super::fiber::FiberseqData;
 use super::*;
-use rayon::{current_num_threads, prelude::*};
+use rayon::prelude::*;
 use rust_htslib::{bam, bam::HeaderView, bam::Read};
 
 pub fn process_bam_chunk(
@@ -91,8 +91,7 @@ pub fn extract_contained(bam: &mut bam::Reader, mut out_files: FiberOut) {
 
     // process bam in chunks
     // keeps mem pretty low, about 1GB per thread
-    let chunk_size = current_num_threads() * 500;
-    let bam_chunk_iter = BamChunk::new(bam.records(), chunk_size);
+    let bam_chunk_iter = BamChunk::new(bam.records(), None);
     for chunk in bam_chunk_iter {
         process_bam_chunk(chunk, &mut out_files, &head_view);
     }

src/nucleosomes.rs

@@ -1,6 +1,5 @@
 use super::fiber::FiberseqData;
 use super::*;
-use rayon::current_num_threads;
 use rayon::iter::ParallelIterator;
 use rayon::prelude::IntoParallelRefMutIterator;
 use rust_htslib::{
@@ -298,8 +297,7 @@ pub fn add_nucleosomes_to_bam(
         allowed_m6a_skips,
     };
     // read in bam data
-    let chunk_size = current_num_threads() * 1000;
-    let bam_chunk_iter = BamChunk::new(bam.records(), chunk_size);
+    let bam_chunk_iter = BamChunk::new(bam.records(), None);
 
     // iterate over chunks
     for mut chunk in bam_chunk_iter {

src/predict_m6a/mod.rs

@@ -5,8 +5,8 @@ use bio_io;
 use nucleosomes;
 use ordered_float::OrderedFloat;
 use rayon::iter::ParallelIterator;
+use rayon::prelude::IndexedParallelIterator;
 use rayon::prelude::IntoParallelRefMutIterator;
-use rayon::{current_num_threads, prelude::IndexedParallelIterator};
 use rust_htslib::{
     bam,
     bam::record::{Aux, AuxArray},
@@ -538,8 +538,7 @@ pub fn read_bam_into_fiberdata(
     predict_options: &PredictOptions,
 ) {
     // read in bam data
-    let chunk_size = current_num_threads() * 500;
-    let bam_chunk_iter = BamChunk::new(bam.records(), chunk_size);
+    let bam_chunk_iter = BamChunk::new(bam.records(), None);
 
     // iterate over chunks
     let mut total_read = 0;

src/strip_basemods.rs

@@ -1,15 +1,13 @@
 use super::basemods;
 use super::*;
-use rayon::current_num_threads;
 use rayon::iter::ParallelIterator;
 use rayon::prelude::IntoParallelRefMutIterator;
 use rust_htslib::bam::Record;
 use rust_htslib::{bam, bam::Read};
 
 pub fn strip_base_mods(bam: &mut bam::Reader, out: &mut bam::Writer, basemod: &str) {
     // read in bam data
-    let chunk_size = current_num_threads() * 500;
-    let bam_chunk_iter = BamChunk::new(bam.records(), chunk_size);
+    let bam_chunk_iter = BamChunk::new(bam.records(), None);
 
     // iterate over chunks
     for mut chunk in bam_chunk_iter {