feat: Add fas2one and fqs2one commands for batch file conversion

crates/deepbiop-cli/src/cli.rs

@@ -16,8 +16,12 @@ pub use extractfq::*;
 pub mod extractfa;
 pub use extractfa::*;
 
-use anyhow::Result;
+pub mod fqs2one;
+pub use fqs2one::*;
+pub mod fas2one;
+pub use fas2one::*;
 
+use anyhow::Result;
 // Set up threads only once, using the common_opts from the top-level Cli struct
 pub fn set_up_threads(threads: Option<usize>) -> Result<()> {
     log::info!("Threads number: {:?}", threads.unwrap());

crates/deepbiop-cli/src/cli/extractfq.rs

@@ -5,13 +5,12 @@ use bstr::BString;
 use clap::Parser;
 use deepbiop_fq as fq;
 
+use super::set_up_threads;
+use log::info;
 use std::io::BufRead;
 use std::io::BufReader;
 use std::path::{Path, PathBuf};
 
-use super::set_up_threads;
-use log::info;
-
 #[derive(Debug, Parser)]
 pub struct ExtractFq {
     /// path to the bam file

crates/deepbiop-cli/src/cli/fas2one.rs

@@ -0,0 +1,29 @@
+use anyhow::Result;
+use clap::Parser;
+use deepbiop_fa as fa;
+
+use std::path::PathBuf;
+
+use super::set_up_threads;
+
+#[derive(Debug, Parser)]
+pub struct FasToOne {
+    /// path to the fa file
+    #[arg(value_name = "fas", action=clap::ArgAction::Append)]
+    fas: Vec<PathBuf>,
+
+    /// output bgzip compressed file
+    #[arg(long, value_name = "output")]
+    output: PathBuf,
+
+    #[arg(short, long, default_value = "2")]
+    threads: Option<usize>,
+}
+
+impl FasToOne {
+    pub fn run(&self) -> Result<()> {
+        set_up_threads(self.threads)?;
+        fa::io::convert_multiple_fas_to_one_bgzip_fa(&self.fas, &self.output, true)?;
+        Ok(())
+    }
+}

crates/deepbiop-cli/src/cli/fq2parquet.rs

@@ -26,7 +26,7 @@ pub struct FqToParquet {
     chunk_size: usize,
 
     /// result path
-    #[arg(long, value_name = "result")]
+    #[arg(long, value_name = "output")]
     output: Option<PathBuf>,
 
     /// threads number

crates/deepbiop-cli/src/cli/fqs2one.rs

@@ -0,0 +1,29 @@
+use anyhow::Result;
+use clap::Parser;
+use deepbiop_fq as fq;
+
+use std::path::PathBuf;
+
+use super::set_up_threads;
+
+#[derive(Debug, Parser)]
+pub struct FqsToOne {
+    /// path to the fq file
+    #[arg(value_name = "fqs", action=clap::ArgAction::Append)]
+    fqs: Vec<PathBuf>,
+
+    /// output bgzip compressed file
+    #[arg(long, value_name = "output")]
+    output: PathBuf,
+
+    #[arg(short, long, default_value = "2")]
+    threads: Option<usize>,
+}
+
+impl FqsToOne {
+    pub fn run(&self) -> Result<()> {
+        set_up_threads(self.threads)?;
+        fq::io::convert_multiple_fqs_to_one_bgzip_fq(&self.fqs, &self.output, true)?;
+        Ok(())
+    }
+}

crates/deepbiop-cli/src/main.rs

@@ -51,6 +51,12 @@ pub enum Commands {
 
     /// Extract fasta reads from a fasta file.
     ExtractFa(cli::ExtractFa),
+
+    /// Multiple Fastqs to one Fastq conversion.
+    FqsToOne(cli::FqsToOne),
+
+    /// Multiple Fastas to one Fasta conversion.
+    FasToOne(cli::FasToOne),
 }
 
 impl Display for Commands {
@@ -64,6 +70,8 @@ impl Display for Commands {
             Commands::ExtractFq(_) => write!(f, "extractfq"),
             Commands::ExtractFa(_) => write!(f, "extractfa"),
             Commands::FqToParquet(_) => write!(f, "fq2parquet"),
+            Commands::FqsToOne(_) => write!(f, "fqs2one"),
+            Commands::FasToOne(_) => write!(f, "fas2one"),
         }
     }
 }
@@ -132,6 +140,14 @@ fn main() -> Result<()> {
             extractfa.run().unwrap();
         }
 
+        Some(Commands::FqsToOne(fqs2one)) => {
+            fqs2one.run().unwrap();
+        }
+
+        Some(Commands::FasToOne(fas2one)) => {
+            fas2one.run().unwrap();
+        }
+
         None => {
             println!("No command provided!");
         }

crates/deepbiop-fa/src/io.rs

@@ -116,27 +116,7 @@ pub fn write_bzip_fa_parallel_for_noodle_record(
     Ok(())
 }
 
-pub fn convert_multiple_fas_to_one_zip_fa<P: AsRef<Path>>(
-    paths: &[PathBuf],
-    result_path: P,
-    parallel: bool,
-) -> Result<()> {
-    let records = if parallel {
-        paths
-            .par_iter()
-            .flat_map(|path| read_noodle_records(path).unwrap())
-            .collect::<Vec<FastaRecord>>()
-    } else {
-        paths
-            .iter()
-            .flat_map(|path| read_noodle_records(path).unwrap())
-            .collect::<Vec<FastaRecord>>()
-    };
-    write_bzip_fa_parallel_for_noodle_record(&records, result_path.as_ref().to_path_buf(), None)?;
-    Ok(())
-}
-
-pub fn convert_multiple_zip_fas_to_one_zip_fa<P: AsRef<Path>>(
+pub fn convert_multiple_fas_to_one_bgzip_fa<P: AsRef<Path>>(
     paths: &[PathBuf],
     result_path: P,
     parallel: bool,

crates/deepbiop-fa/src/python.rs

@@ -176,7 +176,7 @@ fn convert_multiple_fas_to_one_fa(
     let is_zip = paths[0].extension().unwrap() == "gz";
 
     if is_zip {
-        io::convert_multiple_fas_to_one_zip_fa(&paths, result_path, parallel)?;
+        io::convert_multiple_fas_to_one_bzip_fa(&paths, result_path, parallel)?;
     } else {
         io::convert_multiple_zip_fas_to_one_zip_fa(&paths, result_path, parallel)?;
     }

crates/deepbiop-fq/src/io.rs

@@ -100,26 +100,6 @@ pub fn write_bgzip_fq_parallel_for_noodle_record(
     Ok(())
 }
 
-pub fn convert_multiple_bgzip_fqs_to_one_bgzip_fq<P: AsRef<Path>>(
-    paths: &[PathBuf],
-    result_path: P,
-    parallel: bool,
-) -> Result<()> {
-    let records = if parallel {
-        paths
-            .par_iter()
-            .flat_map(|path| read_noodle_records(path).unwrap())
-            .collect::<Vec<FastqRecord>>()
-    } else {
-        paths
-            .iter()
-            .flat_map(|path| read_noodle_records(path).unwrap())
-            .collect::<Vec<FastqRecord>>()
-    };
-    write_bgzip_fq_parallel_for_noodle_record(&records, result_path.as_ref().to_path_buf(), None)?;
-    Ok(())
-}
-
 pub fn convert_multiple_fqs_to_one_bgzip_fq<P: AsRef<Path>>(
     paths: &[PathBuf],
     result_path: P,

crates/deepbiop-fq/src/python.rs

@@ -209,7 +209,7 @@ fn convert_multiple_fqs_to_one_fq(
     if is_zip {
         io::convert_multiple_fqs_to_one_bgzip_fq(&paths, result_path, parallel)?;
     } else {
-        io::convert_multiple_bgzip_fqs_to_one_bgzip_fq(&paths, result_path, parallel)?;
+        io::convert_multiple_fqs_to_one_bgzip_fq(&paths, result_path, parallel)?;
     }
 
     Ok(())