version 0.3.14

sharkLoc · Apr 3, 2024 · 1dc4666 · 1dc4666
1 parent 7f8572f
commit 1dc4666
Show file tree

Hide file tree

Showing 8 changed files with 203 additions and 138 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "fqkit"
-version = "0.3.13"
+version = "0.3.14"
 edition = "2021"
 authors = ["sharkLoc <mmtinfo@163.com>"]
 rust-version = "1.65.0"

diff --git a/README.md b/README.md
@@ -40,7 +40,7 @@ cargo install --git https://github.com/sharkLoc/fqkit.git
 ```bash
 FqKit -- A simple and cross-platform program for fastq file manipulation
 
-Version: 0.3.13
+Version: 0.3.14
 
 Authors: sharkLoc <mmtinfo@163.com>
 Source code: https://github.com/sharkLoc/fqkit.git

diff --git a/src/command.rs b/src/command.rs
diff --git a/src/main.rs b/src/main.rs
@@ -144,18 +144,18 @@ fn main() -> Result<(), Error> {
         Subcli::select { read1, read2, out1, out2 } => {
             select_pe_fastq(&read1, &read2, &out1, &out2, arg.compression_level)?;
         }
-        Subcli::trim { input, left, right, out } => {
+        Subcli::trim { input, left, right, len, out } => {
             if let Some(input) = input {
                 if let Some(out) = out {
-                    trim_fq(&Some(input.as_str()), left, right, &Some(out.as_str()), arg.compression_level)?;
+                    trim_fq(&Some(input.as_str()), left, right, len, &Some(out.as_str()), arg.compression_level)?;
                 } else {
-                    trim_fq(&Some(input.as_str()), left, right, &None, arg.compression_level)?;
+                    trim_fq(&Some(input.as_str()), left, right, len, &None, arg.compression_level)?;
                 }
             } else {
                 if let Some(out) = out {
-                    trim_fq(&None, left, right, &Some(out.as_str()), arg.compression_level)?;
+                    trim_fq(&None, left, right, len, &Some(out.as_str()), arg.compression_level)?;
                 } else {
-                    trim_fq(&None, left, right, &None, arg.compression_level)?;
+                    trim_fq(&None, left, right, len, &None, arg.compression_level)?;
                 }
             }
         }
@@ -367,33 +367,35 @@ fn main() -> Result<(), Error> {
         Subcli::concat { read1, read2, out1, out2 } => {
             concat_fqstq_lane(&read1, &read2, &out1, &out2, arg.compression_level)?;
         }
-        Subcli::remove { input, out, name , save} => {
+        Subcli::remove { input, out, name , save, rm} => {
             if let Some(input) = input {
                 if let Some(out) = out {
-                    remove_read(&Some(&input), &Some(&out) ,&name, &save, arg.compression_level)?;
+                    remove_read(&Some(&input), &Some(&out) ,&name, &save, rm, arg.compression_level)?;
                 } else {
-                    remove_read(&Some(&input), &None ,&name, &save, arg.compression_level)?;
+                    remove_read(&Some(&input), &None ,&name, &save, rm, arg.compression_level)?;
                 }
             } else {
                 if let Some(out) = out {
-                    remove_read(&None, &Some(&out) ,&name, &save, arg.compression_level)?;
+                    remove_read(&None, &Some(&out) ,&name, &save, rm, arg.compression_level)?;
                 } else {
-                    remove_read(&None, &None ,&name, &save, arg.compression_level)?;
+                    remove_read(&None, &None ,&name, &save, rm, arg.compression_level)?;
                 }
             }
         }
-        Subcli::rename { input, keep, prefix, output } => {
+        Subcli::rename { input, keep, prefix, label, before, output } => {
+            let x = label.unwrap_or_default(); 
+            let label = Some(x.as_str());
             if let Some(input) =input {
                 if let Some(output) = output {
-                    rename_fastq(&Some(&input), keep, prefix, &Some(&output), arg.compression_level)?;
+                    rename_fastq(&Some(&input), keep, prefix, label, before, &Some(&output), arg.compression_level)?;
                 } else {
-                    rename_fastq(&Some(&input), keep, prefix, &None, arg.compression_level)?;
+                    rename_fastq(&Some(&input), keep, prefix, label, before, &None, arg.compression_level)?;
                 }
             } else {
                 if let Some(output) = output {
-                    rename_fastq(&None, keep, prefix, &Some(&output), arg.compression_level)?;
+                    rename_fastq(&None, keep, prefix, label, before, &Some(&output), arg.compression_level)?;
                 } else {
-                    rename_fastq(&None, keep, prefix, &None, arg.compression_level)?;
+                    rename_fastq(&None, keep, prefix, label, before, &None, arg.compression_level)?;
                 }
             }
         }

diff --git a/src/remove.rs b/src/remove.rs
@@ -10,6 +10,7 @@ pub fn remove_read(
     out: &Option<&str>,
     name: &str,
     save: &str,
+    rm: bool,
     compression_level: u32,
 ) -> Result<(),Error> {
     if let Some(file) = file {
@@ -18,7 +19,9 @@ pub fn remove_read(
         info!("reading reads from stdin");
     }
     info!("reading reads id form file: {}", name);
-    info!("removed reads in file: {}", save);
+    if !rm {
+        info!("removed reads in file: {}", save);
+    }
     let start = Instant::now();
 
     let mut ids = vec![];
@@ -34,16 +37,26 @@ pub fn remove_read(
 
     let fq_reader = fastq::Reader::new(file_reader(file)?);
     let mut fq_writer = fastq::Writer::new(file_writer(out, compression_level)?);
-    let mut rm_writer = fastq::Writer::new(file_writer(&Some(&save), compression_level)?);
-    for rec in fq_reader.records().flatten() {
-        if !ids.contains(&rec.id().to_string()) {
-            fq_writer.write(rec.id(), rec.desc(), rec.seq(), rec.qual())?;
-        } else {
-            rm_writer.write(rec.id(), rec.desc(), rec.seq(), rec.qual())?;
-        }    
+
+    if rm {
+        for rec in fq_reader.records().flatten() {
+            if !ids.contains(&rec.id().to_string()) {
+                fq_writer.write(rec.id(), rec.desc(), rec.seq(), rec.qual())?;
+            }
+        }
+        fq_writer.flush()?;
+    } else {
+        let mut rm_writer = fastq::Writer::new(file_writer(&Some(&save), compression_level)?);
+        for rec in fq_reader.records().flatten() {
+            if !ids.contains(&rec.id().to_string()) {
+                fq_writer.write(rec.id(), rec.desc(), rec.seq(), rec.qual())?;
+            } else {
+                rm_writer.write(rec.id(), rec.desc(), rec.seq(), rec.qual())?;
+            }    
+        }
+        fq_writer.flush()?;
+        rm_writer.flush()?;
     }
-    fq_writer.flush()?;
-    rm_writer.flush()?;
 
     info!("time elapsed is: {:?}",start.elapsed());
     Ok(())

diff --git a/src/rename.rs b/src/rename.rs
@@ -8,7 +8,9 @@ use crate::utils::*;
 pub fn rename_fastq(
     input: &Option<&str>,
     keep: bool,
-    prefix: Option<String>, //&str,
+    prefix: Option<String>,
+    label: Option<&str>,
+    before: bool,
     output: &Option<&str>,
     compression_level: u32,
 ) -> Result<()> {
@@ -18,15 +20,30 @@ pub fn rename_fastq(
     } else {
         info!("reading from stdin");
     }
-
+  
     let fp = fastq::Reader::new(file_reader(input)?);
     let mut fo = fastq::Writer::new(file_writer(output, compression_level)?);
     let mut n: usize = 0;
 
     if let Some(pre) = prefix {
         for rec in fp.records().flatten() {
             n += 1;
-            let newid = format!("{}{}",pre,n);
+            /*let newid = match label {
+                Some(x) => {
+                    if before { 
+                        format!("{}{}{}",x,pre,n)
+                    } else { 
+                        format!("{}{}{}",pre,x,n)
+                    }
+                },
+                None => { format!("{}{}",pre,n) }
+            };*/
+            let newid = if before {
+                format!("{}{}{}",label.unwrap_or_default(),pre,n)
+            } else {
+                format!("{}{}{}",pre,label.unwrap_or_default(),n)
+            };
+
             let record = if keep { 
                 Record::with_attrs(&newid, rec.desc(), rec.seq(),rec.qual()) 
             } else { 
@@ -38,10 +55,24 @@ pub fn rename_fastq(
     } else {
         for rec in fp.records().flatten() {
             n += 1;
+            /*let newid = if let Some(x) = label {
+                if before { 
+                    format!("{}{}",x,rec.id())
+                } else {
+                    format!("{}{}",rec.id(),x)
+                }
+            } else {
+                format!("{}",rec.id())
+            };*/
+            let newid = if before {
+                format!("{}{}",label.unwrap_or_default(),rec.id())
+            } else {
+                format!("{}{}",rec.id(),label.unwrap_or_default())
+            };
             let record = if keep { 
-                Record::with_attrs(rec.id(), rec.desc(), rec.seq(),rec.qual()) 
+                Record::with_attrs(&newid, rec.desc(), rec.seq(),rec.qual()) 
             } else { 
-                Record::with_attrs(rec.id(), None, rec.seq(),rec.qual())
+                Record::with_attrs(&newid, None, rec.seq(),rec.qual())
             };
             fo.write_record(&record)?;
         }

diff --git a/src/trimfq.rs b/src/trimfq.rs
@@ -9,6 +9,7 @@ pub fn trim_fq(
     file: &Option<&str>,
     left: usize,
     right: usize,
+    len: usize,
     out: &Option<&str>,
     compression_level: u32,
 ) -> Result<()> {
@@ -22,15 +23,20 @@ pub fn trim_fq(
     let length = right + left;
     let fq_reader = fastq::Reader::new(file_reader(file)?);
     let mut fq_writer = fastq::Writer::new(file_writer(out, compression_level)?);
+
     for (idx,rec) in fq_reader.records().flatten().enumerate() {
         let rlen = rec.seq().len();
         if left >= rlen || right>= rlen || length >= rlen {
             warn!("read: {} in order {} is short than {} , skip", rec.id(), idx+1, length);
             continue;
         }
+
         let end = rlen - right;
         let seq = &rec.seq()[left..end];
         let qual = &rec.qual()[left..end];
+        if seq.len() < len {
+            continue;
+        }
         fq_writer.write(rec.id(), rec.desc(), seq, qual)?;
     }
     fq_writer.flush()?;