feat: Add Gap attributes for gff

CHANGELOG.md

@@ -4,7 +4,8 @@
 
 ### Added
 
-- Add a Cigar format
+- Add support of Cigar format
+- Add support of Gff format
 
 ### Changed
 

Cargo.toml

@@ -15,13 +15,14 @@ license-file = "LICENSE"
 
 
 [features]
+cigar    = []
 fasta    = []
 fastq    = []
+gff      = []
 quality  = []
 sequence = []
 vcf      = []
-gff      = []
-cigar    = []
+
 
 [dependencies]
 rand           = { version = "0.8" }

src/format.rs

@@ -5,26 +5,26 @@
 /* crates use */
 
 /* module declaration */
+#[cfg(feature = "cigar")]
+pub mod cigar;
+
 #[cfg(feature = "fasta")]
 pub mod fasta;
 
 #[cfg(feature = "fastq")]
 pub mod fastq;
 
-#[cfg(feature = "vcf")]
-pub mod vcf;
+#[cfg(feature = "gff")]
+pub mod gff;
 
 #[cfg(feature = "sequence")]
 pub mod sequence;
 
 #[cfg(feature = "quality")]
 pub mod quality;
 
-#[cfg(feature = "gff")]
-pub mod gff;
-
-#[cfg(feature = "cigar")]
-pub mod cigar;
+#[cfg(feature = "vcf")]
+pub mod vcf;
 
 /* projet use */
 use crate::error;

src/format/cigar.rs

@@ -50,15 +50,11 @@ impl format::Format for Cigar {
     ) -> error::Result<()> {
         let mut len = 0;
         while len < self.length {
-            dbg!(self.length);
-            dbg!(len);
-            dbg!(1..(self.length - len) as usize);
             let size = if self.length - len > 1 {
                 rng.gen_range::<usize, core::ops::Range<usize>>(1..(self.length - len) as usize)
             } else {
                 1
             };
-            dbg!(size);
 
             let letter = if self.alphabet_weights.is_empty() {
                 self.alphabet.generate(rng, 1)

src/format/gff.rs

@@ -1,4 +1,4 @@
-//! GFF2 format
+//! GFF3 format
 
 /* std use */
 
@@ -15,112 +15,132 @@ use crate::values::Generate as _;
 use crate::values::Get as _;
 
 /// Struct to generate gff record
-#[derive(derive_builder::Builder)]
-#[builder(pattern = "owned")]
+#[derive(typed_builder::TypedBuilder)]
 pub struct Gff {
     /// Chromosome
-    #[builder(default = "values::Chromosomes::Default")]
+    #[builder(default = values::Chromosomes::Default)]
     contigs: values::Chromosomes,
 
     /// Feature
-    #[builder(default = "values::GffFeature::All")]
+    #[builder(default = values::GffFeature::All)]
     features: values::GffFeature,
 
     /// Position
-    #[builder(default = "values::Integer::Position")]
+    #[builder(default = values::Integer::Position)]
     position: values::Integer,
 
     /// Feature length
-    #[builder(default = "values::Integer::UserDefine(1..100_000)")]
+    #[builder(default = values::Integer::UserDefine(1..100_000))]
     length: values::Integer,
 
     /// Score
-    #[builder(default = "values::Float::Default")]
+    #[builder(default = values::Float::Default)]
     score: values::Float,
 
     /// Strand
-    #[builder(default = "values::Strand::All")]
+    #[builder(default = values::Strand::All)]
     strand: values::Strand,
 
     /// Phase
-    #[builder(default = "values::GffPhase::All")]
+    #[builder(default = values::GffPhase::All)]
     phase: values::GffPhase,
 
     /// Id
-    #[builder(default = "values::Alphabet::A2z")]
+    #[builder(default = values::Alphabet::A2z)]
     id: values::Alphabet,
 
     /// Length of id
-    #[builder(default = "10")]
+    #[builder(default = 10)]
     id_len: usize,
 
     /// Id prefix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     id_prefix: Vec<u8>,
 
     /// Id suffix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     id_suffix: Vec<u8>,
 
     /// Name
-    #[builder(default = "values::Alphabet::Lower")]
+    #[builder(default = values::Alphabet::Lower)]
     name: values::Alphabet,
 
     /// Length of name
-    #[builder(default = "10")]
+    #[builder(default = 10)]
     name_len: usize,
 
     /// Name prefix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     name_prefix: Vec<u8>,
 
     /// Name suffix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     name_suffix: Vec<u8>,
 
     /// Alias
-    #[builder(default = "values::Alphabet::A2z")]
+    #[builder(default = values::Alphabet::A2z)]
     alias: values::Alphabet,
 
     /// Length of alias
-    #[builder(default = "10")]
+    #[builder(default = 10)]
     alias_len: usize,
 
     /// Alias prefix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     alias_prefix: Vec<u8>,
 
     /// Alias suffix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     alias_suffix: Vec<u8>,
 
     /// Parent
-    #[builder(default = "values::Alphabet::A2z")]
+    #[builder(default = values::Alphabet::A2z)]
     parent: values::Alphabet,
 
     /// Length of parent
-    #[builder(default = "10")]
+    #[builder(default = 10)]
     parent_len: usize,
 
     /// Parent prefix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     parent_prefix: Vec<u8>,
 
     /// Parent suffix
-    #[builder(default = "b\"\".to_vec()")]
+    #[builder(default = b"".to_vec())]
     parent_suffix: Vec<u8>,
 }
 
 impl Gff {
-    /// Create a GffBuilder
-    pub fn builder() -> GffBuilder {
-        GffBuilder::default()
+    fn produce_gap_value(rng: &mut rand::rngs::StdRng, length: u64) -> error::Result<Vec<u8>> {
+        let mut output = Vec::new();
+        let mut lengths = Vec::new();
+        let mut len = 0;
+        while len < length {
+            let size = if length - len > 1 {
+                rng.gen_range::<usize, core::ops::Range<usize>>(1..(length - len) as usize)
+            } else {
+                1
+            };
+
+            lengths.push(size);
+            len += size as u64;
+        }
+
+        for len in lengths {
+            let letter = values::Cigar::Gff.generate(rng, 1)?;
+            output.extend(letter);
+            output.extend(len.to_string().as_bytes().to_vec());
+            output.push(b' ');
+        }
+        output.pop();
+
+        Ok(output)
     }
 }
 
 impl core::default::Default for Gff {
     fn default() -> Self {
-        GffBuilder::default().build().unwrap() // it's default no error
+        Gff::builder().build()
     }
 }
 
@@ -219,6 +239,11 @@ impl format::Format for Gff {
         output.write_all(&self.parent_prefix)?;
         output.write_all(&self.parent.generate(rng, self.parent_len)?)?;
         output.write_all(&self.parent_suffix)?;
+        output.write_all(b";")?;
+
+        // gap
+        output.write_all(b"Gap=")?;
+        output.write_all(&Gff::produce_gap_value(rng, (end - start) as u64)?)?;
 
         Ok(())
     }
@@ -233,14 +258,14 @@ mod tests {
     use super::format::Format as _;
     use super::*;
 
-    const TRUTH: &[u8] = b"YAR028W\tbiotest\texon\t6057\t6155\t9.429573\t.\t0\tID=[tBTlDDl[M;Name=emxuzgaghm;Alias=s^[teLMir[;Parent=gMDhw\\voCG
-YAR028W\tbiotest\trepeat\t4903\t4948\t7.9373302\t+\t0\tID=gQouVGn`Jw;Name=qjbbjlzxpz;Alias=Any[_POshs;Parent=qbSjAdbZcR
-YAR028W\tbiotest\trepeat\t5211\t5297\t3.2389307\t+\t0\tID=tSvzMeTjon;Name=ljdusfsrcu;Alias=tTH\\QXXOiA;Parent=LJLnuPtf`S
-YAR028W\tbiotest\trepeat\t5617\t5673\t3.21298\t+\t2\tID=vYGCzkT\\Wk;Name=nwbrlpbpvm;Alias=tIGEbcnVWJ;Parent=VaDBnQSHYN
-X\tbiotest\ttranscript\t5944\t6040\t9.520424\t+\t2\tID=`NnOG[K`QK;Name=rrilpylxga;Alias=MyDqpgZliS;Parent=mUzRvGGXBg
+    const TRUTH: &[u8] = b"YAR028W\tbiotest\texon\t6057\t6155\t9.429573\t.\t0\tID=[tBTlDDl[M;Name=emxuzgaghm;Alias=s^[teLMir[;Parent=gMDhw\\voCG;Gap=M48 D11 R26 F10 M1 D1 D1
+93\tbiotest\texon\t8323\t8381\t3.2013047\t-\t.\tID=dbZcRFrrQ_;Name=jwinicfqqi;Alias=jonYVInjLI;Parent=i`oWogntTH;Gap=I28 R12 F5 I1 R3 D8 I1
+X\tbiotest\texon\t9176\t9219\t1.0694146\t.\t.\tID=zkT\\Wk_sGD;Name=rlpbpvmdcp;Alias=nVWJVaDBnQ;Parent=SHYNm[QBCg;Gap=F24 R15 F2 I1 D1
+ENA|LT795502|LT795502.1\tbiotest\tgene\t2073\t2169\t0.5253875\t-\t2\tID=gZliSmUzRv;Name=ccdkarvolo;Alias=Bw_ZxxkAFA;Parent=[o`OIdJgjZ;Gap=R31 F13 I47 D4 F1
+ENA|LT795502|LT795502.1\tbiotest\ttranscript\t3919\t3944\t9.702128\t.\t0\tID=jBlBKigqzn;Name=gultrkslsv;Alias=\\RlwOmAiZP;Parent=wyAsKBssXJ;Gap=R6 R4 D2 D10 F2 I1
 ";
 
-    const DEFAULT: &[u8] = b"YAR028W\tbiotest\texon\t1133862760\t1133889429\t21.144531\t.\t0\tID=[tBTlDDl[M;Name=emxuzgaghm;Alias=s^[teLMir[;Parent=gMDhw\\voCG";
+    const DEFAULT: &[u8] = b"YAR028W\tbiotest\texon\t1133862760\t1133889429\t21.144531\t.\t0\tID=[tBTlDDl[M;Name=emxuzgaghm;Alias=s^[teLMir[;Parent=gMDhw\\voCG;Gap=D21168 D1146 R2911 I1127 D50 I96 R103 R1 M46 I7 F1 F1 F7 R1 D2 F1 F1";
 
     #[test]
     fn default() -> error::Result<()> {
@@ -265,12 +290,11 @@ X\tbiotest\ttranscript\t5944\t6040\t9.520424\t+\t2\tID=`NnOG[K`QK;Name=rrilpylxg
             .position(values::Integer::UserDefine(0..10_000))
             .length(values::Integer::UserDefine(2..100))
             .score(values::Float::UserDefine(0.0..10.0))
-            .build()?;
+            .build();
 
         generator.record(&mut output, &mut rng)?;
 
-        println!("{:?} {}", String::from_utf8(output.to_vec()), output.len());
-        assert_eq!(output, TRUTH.to_vec()[..108]);
+        assert_eq!(output, TRUTH.to_vec()[..137]);
 
         Ok(())
     }
@@ -284,7 +308,7 @@ X\tbiotest\ttranscript\t5944\t6040\t9.520424\t+\t2\tID=`NnOG[K`QK;Name=rrilpylxg
             .position(values::Integer::UserDefine(0..10_000))
             .length(values::Integer::UserDefine(2..100))
             .score(values::Float::UserDefine(0.0..10.0))
-            .build()?;
+            .build();
 
         generator.records(&mut output, &mut rng, 5)?;
 
@@ -306,7 +330,7 @@ X\tbiotest\ttranscript\t5944\t6040\t9.520424\t+\t2\tID=`NnOG[K`QK;Name=rrilpylxg
             .position(values::Integer::UserDefine(0..10_000))
             .length(values::Integer::UserDefine(2..100))
             .score(values::Float::UserDefine(0.0..10.0))
-            .build()?;
+            .build();
 
         generator.create(&temp_file, &mut rng, 5)?;
 

src/values.rs

@@ -361,6 +361,8 @@ impl core::convert::AsRef<[&'static [u8]]> for Strand {
         }
     }
 }
+
+#[derive(Default)]
 /// Possible cigar alphabet
 pub enum Cigar {
     #[default]
@@ -380,6 +382,8 @@ impl core::convert::AsRef<[u8]> for Cigar {
     }
 }
 
+impl Generate for Cigar {}
+
 #[derive(Debug, Clone, Default)]
 /// Possible value for frame
 pub enum GffFeature {
@@ -428,8 +432,6 @@ impl core::convert::AsRef<[&'static [u8]]> for GffPhase {
     }
 }
 
-impl Generate for Cigar {}
-
 #[cfg(test)]
 mod tests {
     /* project use */
@@ -659,6 +661,7 @@ mod tests {
         );
     }
 
+    #[test]
     fn cigar() -> error::Result<()> {
         assert_eq!(Cigar::Sam.as_ref(), constants::CIGAR_SAM);
         assert_eq!(Cigar::Gff.as_ref(), constants::CIGAR_GFF);