feat: add gc content, read length and N base count

in output for subcli flat

src/cli/flatten.rs

@@ -9,6 +9,9 @@ pub fn flatten_fq(
     out: Option<&String>,
     flag: u8,
     sep: char,
+    gap: bool,
+    len: bool,
+    gc: bool,
     compression_level: u32,
 ) -> Result<(), Error> {
     let start = Instant::now();
@@ -37,11 +40,30 @@ pub fn flatten_fq(
 
     for rec in fq_reader.records().flatten() {
         let read = vec![rec.id().as_bytes(), rec.seq(), "+".as_bytes(), rec.qual()];
-        let res = fields.iter().map(|idx| read[*idx]).collect::<Vec<&[u8]>>();
+        let res = fields
+            .iter()
+            .map(|idx| read[*idx])
+            .collect::<Vec<&[u8]>>();
+
         let mut out = Vec::new();
         for x in res {
             out.push(std::str::from_utf8(x)?.to_string());
         }
+        if gap {
+            out.push(rec.seq().iter().filter(|x| *x == &b'N').count().to_string());
+        }
+        if len {
+            out.push(rec.seq().len().to_string());
+        }
+        if gc {
+            let gc_count = rec
+            .seq()
+            .iter()
+            .filter(|x| *x == &b'G' || *x == &b'C')
+            .count();
+            let gc_ratio = format!("{:.2}",gc_count as f64 / rec.seq().len() as f64 * 100.0);
+            out.push(gc_ratio);
+        }
         out_writer.write_all(out.join(sep.to_string().as_str()).as_bytes())?;
         out_writer.write_all("\n".as_bytes())?;
     }

src/command.rs

@@ -435,6 +435,15 @@ pub enum Subcli {
         /// output seprater, can be ",",  ";",
         #[arg(short = 's', long = "sep", default_value_t = '\t', value_name = "CHAR")]
         sep: char,
+        /// if specified, add N base count in output
+        #[arg(short = 'n', long = "gap-n", help_heading = Some("FLAGS"))]
+        gap: bool,
+        /// if specified, add read length in output
+        #[arg(short = 'l', long = "length", help_heading = Some("FLAGS"))]
+        len: bool,
+        /// if specified, add GC content(%) in output
+        #[arg(short = 'g', long = "gc-content", help_heading = Some("FLAGS"))]
+        gc: bool,
         /// output file name or write to stdout, file ending in .gz/.bz2/.xz will be compressed automatically
         #[arg(short = 'o', long = "out", value_name = "FILE")]
         out: Option<String>,

src/main.rs

@@ -170,13 +170,19 @@ fn main() -> Result<(), Error> {
             input,
             flag,
             sep,
+            gap,
+            len,
+            gc,
             out,
         } => {
             flatten_fq(
                 input.as_ref(),
                 out.as_ref(),
                 flag,
                 sep,
+                gap,
+                len,
+                gc,
                 arg.compression_level,
             )?;
         }