Merge pull request #19 from cschin/main

refactoring the python lib
GeneDx · May 5, 2023 · bd4efd3 · bd4efd3
2 parents 0d5b135 + 2af0ef3
commit bd4efd3
Show file tree

Hide file tree

Showing 15 changed files with 447 additions and 797 deletions.
diff --git a/build.sh b/build.sh
@@ -1,6 +1,6 @@
-pushd WFA2-lib
-make all
-popd
+#pushd WFA2-lib
+#make all
+#popd
 
 rustup default stable
 cargo build -p pgr-db --release

diff --git a/pgr-bin/Cargo.toml b/pgr-bin/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "pgr-bin"
-version = "0.5.0"
+version = "0.5.1"
 edition = "2021"
 authors = ["Jason Chin <cschin@infoecho.net>"]
 

diff --git a/pgr-bin/src/bin/pgr-compare-cov.rs b/pgr-bin/src/bin/pgr-compare-cov.rs
@@ -3,14 +3,14 @@ const VERSION_STRING: &str = env!("VERSION_STRING");
 //use std::path::PathBuf;
 use clap::{self, CommandFactory, Parser};
 
-use pgr_bin::{pair_shmmrs, sequence_to_shmmrs, SeqIndexDB, ShmmrSpec};
+use pgr_db::ext::{pair_shmmrs, sequence_to_shmmrs, SeqIndexDB, ShmmrSpec};
+use rayon::prelude::*;
 use rustc_hash::FxHashSet;
 use std::{
     fs::File,
     io::{BufRead, BufReader, BufWriter, Write},
     path::Path,
 };
-use rayon::prelude::*;
 
 /// Compare SHIMMER pair count in two input sequence files
 #[derive(Parser, Debug)]
@@ -320,7 +320,7 @@ fn generate_bed_graph_from_sdb(args: &CmdOptions, input_type: &str) {
     } else {
         panic!("input type has to be specified  AGC or FRG backends")
     };
-    
+
     #[cfg(not(feature = "with_agc"))]
     if input_type == "FRG" {
         let _ = seq_index_db.load_from_frg_index(args.frg_idx_prefix.as_ref().unwrap().clone());

diff --git a/pgr-bin/src/bin/pgr-compare-cov2.rs b/pgr-bin/src/bin/pgr-compare-cov2.rs
@@ -3,7 +3,8 @@ const VERSION_STRING: &str = env!("VERSION_STRING");
 //use std::path::PathBuf;
 use clap::{self, CommandFactory, Parser};
 
-use pgr_bin::{pair_shmmrs, sequence_to_shmmrs, SeqIndexDB};
+use pgr_db::ext::{pair_shmmrs, sequence_to_shmmrs, SeqIndexDB};
+use pgr_db::seq_db::{get_shmmr_matches_from_mmap_file, ShmmrPair};
 use rayon::prelude::*;
 use rustc_hash::FxHashSet;
 use std::{
@@ -179,7 +180,32 @@ fn generate_bed_graph_from_sdb(args: &CmdOptions, input_type: &str) {
                     }
                 });
 
-            let frag_map = seq_index_db.get_shmmr_map_internal().unwrap();
+            //let frag_map = seq_index_db.get_shmmr_map_internal().unwrap();
+
+            let get_shmmr_matches = |smps: ShmmrPair| {
+                #[cfg(feature = "with_agc")]
+                if input_type == "AGC" {
+                    get_shmmr_matches_from_mmap_file(
+                        &seq_index_db.agc_db.as_ref().unwrap().frag_location_map,
+                        smps,
+                        &seq_index_db.agc_db.as_ref().unwrap().frag_map_file,
+                    )
+                } else {
+                    get_shmmr_matches_from_mmap_file(
+                        &seq_index_db.frg_db.as_ref().unwrap().frag_location_map,
+                        smps,
+                        &seq_index_db.frg_db.as_ref().unwrap().frag_map_file,
+                    )
+                }
+                #[cfg(not(feature = "with_agc"))]
+                {
+                    get_shmmr_matches_from_mmap_file(
+                        &seq_index_db.frg_db.as_ref().unwrap().frag_location_map,
+                        smps,
+                        &seq_index_db.frg_db.as_ref().unwrap().frag_map_file,
+                    )
+                }
+            };
             let mut output_bedgraph_file0 = BufWriter::new(
                 File::create(Path::new(&prefix).with_extension("0.bedgraph"))
                     .expect("can't create the output file"),
@@ -217,7 +243,8 @@ fn generate_bed_graph_from_sdb(args: &CmdOptions, input_type: &str) {
                         } else {
                             (s1, s0, p0, p1, 1_u8)
                         };
-                        let (c0, c1) = if let Some(hits) = frag_map.get(&(k.0, k.1)) {
+                        let (c0, c1) = {
+                            let hits = get_shmmr_matches((k.0, k.1));
                             let mut c0 = 0_usize;
                             let mut c1 = 0_usize;
                             hits.iter().for_each(|v| {
@@ -230,8 +257,6 @@ fn generate_bed_graph_from_sdb(args: &CmdOptions, input_type: &str) {
                                 }
                             });
                             (c0, c1)
-                        } else {
-                            (0, 0)
                         };
                         assert!(c0 > 0);
                         let r = c1 as f32 / c0 as f32;
@@ -273,7 +298,8 @@ fn generate_bed_graph_from_sdb(args: &CmdOptions, input_type: &str) {
                         } else {
                             (s1, s0, p0, p1, 1_u8)
                         };
-                        let (c0, c1) = if let Some(hits) = frag_map.get(&(k.0, k.1)) {
+                        let (c0, c1) = {
+                            let hits = get_shmmr_matches((k.0, k.1));
                             let mut c0 = 0_usize;
                             let mut c1 = 0_usize;
                             hits.iter().for_each(|v| {
@@ -286,8 +312,6 @@ fn generate_bed_graph_from_sdb(args: &CmdOptions, input_type: &str) {
                                 }
                             });
                             (c0, c1)
-                        } else {
-                            (0, 0)
                         };
                         assert!(c1 > 0);
                         let r = c0 as f32 / c1 as f32;

diff --git a/pgr-bin/src/bin/pgr-fetch-seqs.rs b/pgr-bin/src/bin/pgr-fetch-seqs.rs
@@ -1,6 +1,6 @@
 const VERSION_STRING: &str = env!("VERSION_STRING");
 use clap::{self, CommandFactory, Parser};
-use pgr_bin::SeqIndexDB;
+use pgr_db::ext::SeqIndexDB;
 use pgr_db::fasta_io;
 use std::fs::File;
 use std::io::{self, BufRead, BufReader, BufWriter, Write};
@@ -88,7 +88,7 @@ fn main() -> Result<(), std::io::Error> {
     } else {
         Box::new(io::stdout())
     };
-    
+
     region_file.lines().into_iter().for_each(|line| {
         let line = line.expect("fail to get a line in the region file");
         let fields = line.split('\t').collect::<Vec<&str>>();

diff --git a/pgr-bin/src/bin/pgr-make-frgdb.rs b/pgr-bin/src/bin/pgr-make-frgdb.rs
@@ -3,7 +3,7 @@ const VERSION_STRING: &str = env!("VERSION_STRING");
 //use std::path::PathBuf;
 use clap::{self, CommandFactory, Parser};
 
-use pgr_bin::SeqIndexDB;
+use pgr_db::ext::SeqIndexDB;
 use std::fs::File;
 use std::io::{BufRead, BufReader};
 use std::path::Path;

diff --git a/pgr-bin/src/bin/pgr-pbundle-decomp.rs b/pgr-bin/src/bin/pgr-pbundle-decomp.rs
@@ -1,7 +1,7 @@
 const VERSION_STRING: &str = env!("VERSION_STRING");
 use bincode::config;
 use clap::{self, CommandFactory, Parser};
-use pgr_bin::{
+use pgr_db::ext::{
     get_principal_bundle_decomposition, PrincipalBundlesWithId, SeqIndexDB, VertexToBundleIdMap,
 };
 use rustc_hash::{FxHashMap, FxHashSet};
@@ -200,7 +200,7 @@ fn main() -> Result<(), std::io::Error> {
                     usize,
                     usize,
                     PrincipalBundlesWithId,
-                    pgr_bin::VertexToBundleIdMap,
+                    VertexToBundleIdMap,
                 ),
                 config::Configuration,
             >(&s[..], config)