mapq filtering: Implement it.

Fixes #219.

Suggested by: @SDmetagenomics.

Cargo.toml

@@ -12,6 +12,7 @@ exclude = [ # Max upload is 10MB, as of writing test data was 15MB
   "tests/*",
 ]
 default-run = "coverm"
+edition = "2015"
 
 [dependencies]
 bio = "1.1.0"

src/bam_generator.rs

@@ -486,6 +486,7 @@ pub fn generate_filtered_bam_readers_from_bam_files(
     min_aligned_length_single: u32,
     min_percent_identity_single: f32,
     min_aligned_percent_single: f32,
+    min_mapq: u8,
     min_aligned_length_pair: u32,
     min_percent_identity_pair: f32,
     min_aligned_percent_pair: f32,
@@ -511,6 +512,7 @@ pub fn generate_filtered_bam_readers_from_bam_files(
                 min_aligned_length_single,
                 min_percent_identity_single,
                 min_aligned_percent_single,
+                min_mapq,
                 min_aligned_length_pair,
                 min_percent_identity_pair,
                 min_aligned_percent_pair,
@@ -544,6 +546,7 @@ pub struct StreamingFilteredNamedBamReaderGenerator {
     min_aligned_length_single: u32,
     min_percent_identity_single: f32,
     min_aligned_percent_single: f32,
+    min_mapq: u8,
     min_aligned_length_pair: u32,
     min_percent_identity_pair: f32,
     min_aligned_percent_pair: f32,
@@ -584,6 +587,7 @@ impl NamedBamReaderGenerator<StreamingFilteredNamedBamReader>
             self.min_aligned_length_single,
             self.min_percent_identity_single,
             self.min_aligned_percent_single,
+            self.min_mapq,
             self.min_aligned_length_pair,
             self.min_percent_identity_pair,
             self.min_aligned_percent_pair,
@@ -651,6 +655,7 @@ pub fn generate_filtered_named_bam_readers_from_reads(
     min_aligned_length_single: u32,
     min_percent_identity_single: f32,
     min_aligned_percent_single: f32,
+    min_mapq: u8,
     min_aligned_length_pair: u32,
     min_percent_identity_pair: f32,
     min_aligned_percent_pair: f32,
@@ -682,6 +687,7 @@ pub fn generate_filtered_named_bam_readers_from_reads(
         min_aligned_length_single,
         min_percent_identity_single,
         min_aligned_percent_single,
+        min_mapq,
         min_aligned_length_pair,
         min_percent_identity_pair,
         min_aligned_percent_pair,

src/bin/coverm.rs

@@ -165,6 +165,7 @@ fn main() {
                             filter_params.min_aligned_length_single,
                             filter_params.min_percent_identity_single,
                             filter_params.min_aligned_percent_single,
+                            filter_params.min_mapq,
                             filter_params.min_aligned_length_pair,
                             filter_params.min_percent_identity_pair,
                             filter_params.min_aligned_percent_pair,
@@ -463,6 +464,7 @@ fn main() {
                     filter_params.min_aligned_length_single,
                     filter_params.min_percent_identity_single,
                     filter_params.min_aligned_percent_single,
+                    filter_params.min_mapq,
                     filter_params.min_aligned_length_pair,
                     filter_params.min_percent_identity_pair,
                     filter_params.min_aligned_percent_pair,
@@ -519,6 +521,7 @@ fn main() {
                             filter_params.min_aligned_length_single,
                             filter_params.min_percent_identity_single,
                             filter_params.min_aligned_percent_single,
+                            filter_params.min_mapq,
                             filter_params.min_aligned_length_pair,
                             filter_params.min_percent_identity_pair,
                             filter_params.min_aligned_percent_pair,
@@ -1209,6 +1212,7 @@ struct FilterParameters {
     min_aligned_length_single: u32,
     min_percent_identity_single: f32,
     min_aligned_percent_single: f32,
+    min_mapq: u8, // 255 indicates no filtering
     min_aligned_length_pair: u32,
     min_percent_identity_pair: f32,
     min_aligned_percent_pair: f32,
@@ -1224,6 +1228,7 @@ impl FilterParameters {
             min_aligned_length_single: *m.get_one::<u32>("min-read-aligned-length").unwrap_or(&0),
             min_percent_identity_single: parse_percentage(m, "min-read-percent-identity"),
             min_aligned_percent_single: parse_percentage(m, "min-read-aligned-percent"),
+            min_mapq: *m.get_one::<u8>("min-mapq").unwrap_or(&255),
             min_aligned_length_pair: *m
                 .get_one::<u32>("min-read-aligned-length-pair")
                 .unwrap_or(&0),
@@ -1253,6 +1258,7 @@ impl FilterParameters {
         self.min_percent_identity_single > 0.0
             || self.min_percent_identity_pair > 0.0
             || self.min_aligned_percent_single > 0.0
+            || self.min_mapq < 255
             || self.min_aligned_percent_pair > 0.0
             || self.min_aligned_length_single > 0
             || self.min_aligned_length_pair > 0
@@ -1578,6 +1584,7 @@ fn get_streamed_filtered_bam_readers(
                     filter_params.min_aligned_length_single,
                     filter_params.min_percent_identity_single,
                     filter_params.min_aligned_percent_single,
+                    filter_params.min_mapq,
                     filter_params.min_aligned_length_pair,
                     filter_params.min_percent_identity_pair,
                     filter_params.min_aligned_percent_pair,

src/cli.rs

@@ -175,6 +175,16 @@ fn add_thresholding_options(manual: Manual) -> Manual {
                         default_roff("0")
                     )),
             )
+            .option(Opt::new("INT").long("--min-mapq").help(&format!(
+                "Exclude reads with a mapping quality \
+                below this value. An INT between 0 and 254. \
+                When thresholding pairs, both reads \
+                are excluded if either has a MAPQ below this value. \
+                If 0, \
+                reads with MAPQ of 255 (mapping quality unavailable) \
+                are excluded. {}",
+                default_roff("not set")
+            )))
             .flag(
                 Flag::new()
                     .long("--proper-pairs-only")
@@ -1322,6 +1332,11 @@ Ben J. Woodcroft <benjwoodcroft near gmail.com>
                         .long("min-read-aligned-percent")
                         .value_parser(clap::value_parser!(f32)),
                 )
+                .arg(
+                    Arg::new("min-mapq")
+                        .long("min-mapq")
+                        .value_parser(clap::value_parser!(u8)),
+                )
                 .arg(
                     Arg::new("min-read-aligned-length-pair")
                         .long("min-read-aligned-length-pair")
@@ -1707,6 +1722,11 @@ Ben J. Woodcroft <benjwoodcroft near gmail.com>
                         .long("min-read-aligned-percent")
                         .value_parser(clap::value_parser!(f32)),
                 )
+                .arg(
+                    Arg::new("min-mapq")
+                        .long("min-mapq")
+                        .value_parser(clap::value_parser!(u8)),
+                )
                 .arg(
                     Arg::new("min-read-aligned-length-pair")
                         .long("min-read-aligned-length-pair")
@@ -1845,6 +1865,11 @@ Ben J. Woodcroft <benjwoodcroft near gmail.com>
                         .long("min-read-percent-identity")
                         .value_parser(clap::value_parser!(f32)),
                 )
+                .arg(
+                    Arg::new("min-mapq")
+                        .long("min-mapq")
+                        .value_parser(clap::value_parser!(u8)),
+                )
                 .arg(
                     Arg::new("min-read-aligned-percent")
                         .long("min-read-aligned-percent")

src/contig.rs

@@ -273,7 +273,6 @@ mod tests {
     use mapping_index_maintenance::VanillaIndexStruct;
     use mapping_parameters::*;
     use shard_bam_reader::*;
-    use std::io::Read;
     use std::str;
     use OutputWriter;
 

src/coverage_printer.rs

@@ -544,7 +544,6 @@ pub fn print_dense_cached_coverage_taker(
 mod tests {
     use super::*;
     use std::io::Cursor;
-    use std::io::Read;
     use std::str;
     use OutputWriter;