IBCHgenomic

starting the final compilation

Author: codebiogenomics

README.md

@@ -5,6 +5,30 @@
 - Parallel threaded variant linker.
 - sample vcf file from [vcflib](https://github.com/vcflib/vcflib/blob/master/samples/sample.vcf)
 
+
+```
+specific position annotator for human genomics.
+       ************************************************
+      Gaurav Sablok, IBCH, PAN, Poznan, Poland,
+      https://portal.ichb.pl/laboratory-of-genomics/.
+      Email: [email protected]
+      Prof. Luiza Handschuh
+      Email: [email protected].
+      ************************************************
+
+Usage: varlinker <COMMAND>
+
+Commands:
+  variantlinker   annotate the specific coordinate
+  variantrefanno  extract the annotation of the specific ref allele
+  variantaltanno  extract the annotation of the specific alt allele
+  help            Print this message or the help of the given subcommand(s)
+
+Options:
+  -h, --help     Print help
+  -V, --version  Print version
+```
+
 - Acknowledgements: MOSAIC platform, developed as part of the ECBiG-MOSAIC project (POIR.04.02.00-00-D017/20), co-financed by the European Regional Development Fund (ERDF) under the Smart Growth Operational Programme 2014-2020, Measure 4.2 for the development of modern research infrastructure in the science sector.
 - Project PI and Informal queries: Prof. Luiza Handschuh: [email protected].
 - Code related queries: Dr. Gaurav Sablok: [email protected].

src/args.rs

@@ -25,4 +25,18 @@ pub enum Commands {
         /// variant VCF file
         vcfile: String,
     },
+    /// extract the annotation of the specific ref allele
+    VARIANTREFANNO {
+        /// variant VCF file
+        vcffile: String,
+        /// ref allele
+        refallele: String,
+    },
+    /// extract the annotation of the specific alt allele
+    VARIANTALTANNO {
+        /// variant VCF file
+        vcffile: String,
+        /// alt allele
+        altallel: String,
+    }
 }

src/main.rs

@@ -1,9 +1,15 @@
 mod args;
 mod store;
 mod variantlinker;
+mod varaltannot;
+mod varrefannot;
 use crate::args::CommandParse;
 use crate::args::Commands;
 use clap::Parser;
+use crate::variantlinker::varlinker;
+use crate::varrefannot::varrefanno;
+use crate::varaltannot::varaltanno;
+
 
 /*
  Authom GauravSablok
@@ -16,13 +22,16 @@ use clap::Parser;
 fn main() {
     let argsparse = CommandParse::parse();
     match &argsparse.command {
-        Commands::CUIGENERATE {
-            medgenhpo,
-            medgen_omim,
-            medgenmapping,
-            medgenpubmed,
-        } => {
-            let command = cuiparallel(medgenhpo, medgen_omim, medgenmapping, medgenpubmed).unwrap();
+        Commands::VARIANTLINKER { vcfile } => {
+            let command = varlinker(vcfile).unwrap();
+            println!("The command has been completed:{:?}", command);
+        }
+        Commands::VARIANTALTANNO { vcffile, altallel } => {
+            let command = varaltanno(vcffile, altallel).unwrap();
+            println!("The command has been completed:{:?}", command);
+        }
+        Commands::VARIANTREFANNO { vcffile, refallele } => {
+            let command = varrefanno(vcffile, refallele).unwrap();
             println!("The command has been completed:{:?}", command);
         }
     }

src/store.rs

@@ -12,5 +12,6 @@ pub struct VCF {
 #[derive(Debug, Clone, PartialOrd, PartialEq)]
 
 pub struct GENCODE {
-    pub
+    pub chrom: String,
+    pub typeannotate: String,
 }

src/varaltannot.rs

@@ -0,0 +1,52 @@
+use std::fs::File;
+use std::io::{BufReader, BufRead};
+use std::error::Error;
+use std::process::Command;
+use crate::store::VCF;
+
+
+/*
+ Author Gaurav Sablok
+ Instytut Chemii Bioorganicznej
+ Polskiej Akademii Nauk
+ ul. Noskowskiego 12/14 | 61-704, Poznań
+ Date: 2025-6-9
+*/
+
+pub fn varaltanno(pathfile : &str, variant: &str) -> Result<String, Box<dyn Error>>{
+
+    let _ = Command::new("wegt").
+        arg("https://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_48/gencode.v48.chr_patch_hapl_scaff.annotation.gtf.gz")
+        .output()
+        .expect("command failed");
+    let _ = Command::new("gunzip").arg("gencode.v48.chr_patch_hapl_scaff.annotation.gtf.gz")
+        .output()
+        .expect("command failed");
+    let fileopen = File::open(pathfile).expect("file not present");
+    let fileread = BufReader::new(fileopen);
+    let gtfopen = File::open("gencode.v48.chr_patch_hapl_scaff.annotation.gtf")
+        .expect("file not present");
+    let gtfread = BufReader::new(gtfopen);
+    let exonvector:Vec<String> = Vec::new();
+    let gene: Vec<String> = Vec::new();
+    let mut vcstring_file: Vec<VCF> = Vec::new();
+    for i in fileread.lines(){
+        let linevcf = i.expect("file not present");
+        let linevec:Vec<String> = linevcf.split("\t").
+            map(|x|x.to_string())
+            .collect::<Vec<_>>();
+        vcstring_file.push(VCF{
+                  chrom: linevec[0].to_string(),
+                  pos: linevec[1].parse::<usize>().unwrap(),
+                  id: linevec[2].to_string(),
+                  refnuc: linevec[3].to_string(),
+                  altnuc: linevec[4].to_string(),
+                  qual: linevec[5].to_string(),
+        });
+    }
+
+
+
+
+    Ok("The regions have been annotated".to_string())
+}

src/variantlinker.rs

@@ -6,14 +6,14 @@ use crate::store::VCF;
 
 
 /*
- Authom GauravSablok
+ Author Gaurav Sablok
  Instytut Chemii Bioorganicznej
  Polskiej Akademii Nauk
  ul. Noskowskiego 12/14 | 61-704, Poznań
- Date: 2025-4-8
+ Date: 2025-6-9
 */
 
-pub async fn varlinker(pathfile : &str) -> Result<String, Box<dyn Error>>{
+pub fn varlinker(pathfile : &str) -> Result<String, Box<dyn Error>>{
 
     let _ = Command::new("wegt").
         arg("https://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_48/gencode.v48.chr_patch_hapl_scaff.annotation.gtf.gz")

src/varrefannot.rs

@@ -0,0 +1,52 @@
+use std::fs::File;
+use std::io::{BufReader, BufRead};
+use std::error::Error;
+use std::process::Command;
+use crate::store::VCF;
+
+
+/*
+ Author Gaurav Sablok
+ Instytut Chemii Bioorganicznej
+ Polskiej Akademii Nauk
+ ul. Noskowskiego 12/14 | 61-704, Poznań
+ Date: 2025-6-9
+*/
+
+pub fn varrefanno(pathfile : &str, variant: &str) -> Result<String, Box<dyn Error>>{
+
+    let _ = Command::new("wegt").
+        arg("https://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_48/gencode.v48.chr_patch_hapl_scaff.annotation.gtf.gz")
+        .output()
+        .expect("command failed");
+    let _ = Command::new("gunzip").arg("gencode.v48.chr_patch_hapl_scaff.annotation.gtf.gz")
+        .output()
+        .expect("command failed");
+    let fileopen = File::open(pathfile).expect("file not present");
+    let fileread = BufReader::new(fileopen);
+    let gtfopen = File::open("gencode.v48.chr_patch_hapl_scaff.annotation.gtf")
+        .expect("file not present");
+    let gtfread = BufReader::new(gtfopen);
+    let exonvector:Vec<String> = Vec::new();
+    let gene: Vec<String> = Vec::new();
+    let mut vcstring_file: Vec<VCF> = Vec::new();
+    for i in fileread.lines(){
+        let linevcf = i.expect("file not present");
+        let linevec:Vec<String> = linevcf.split("\t").
+            map(|x|x.to_string())
+            .collect::<Vec<_>>();
+        vcstring_file.push(VCF{
+                  chrom: linevec[0].to_string(),
+                  pos: linevec[1].parse::<usize>().unwrap(),
+                  id: linevec[2].to_string(),
+                  refnuc: linevec[3].to_string(),
+                  altnuc: linevec[4].to_string(),
+                  qual: linevec[5].to_string(),
+        });
+    }
+
+
+
+
+    Ok("The regions have been annotated".to_string())
+}