IBCHGenomic

Finished the first part and writing the last part

Author: codebiogenomics

main.rs

@@ -0,0 +1,93 @@
+use crate::annotation::ontologyannotate;
+use crate::args::CommandParse;
+use crate::args::Commands;
+use crate::clinicvar::clinvarmapper;
+use crate::clinvarlinker::clinvarvcf;
+use crate::medgen::cuiparallel;
+use crate::ncbigeneid::ncbiannotate;
+use crate::omim::omimevidence;
+use crate::phenotype::phenotypeall;
+use crate::databases::databasedownload;
+use clap::Parser;
+
+/*
+ Authom GauravSablok
+ Instytut Chemii Bioorganicznej
+ Polskiej Akademii Nauk
+ ul. Noskowskiego 12/14 | 61-704, Poznań
+ Date: 2025-4-8
+*/
+
+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();
+            println!("The command has been completed:{:?}", command);
+        }
+        Commands::OMIM {
+            omimfile,
+            evidencenumber,
+            hpomapping,
+            hpomedgen,
+        } => {
+            let command = omimevidence(omimfile, evidencenumber, hpomapping, hpomedgen).unwrap();
+            println!("The links for the given evidence are:{:?}", command);
+        }
+        Commands::CLINVAROMIMEVIDENCE {
+            clinvar,
+            medgen,
+            medgenhpo,
+            omim,
+        } => {
+            let command = clinvarmapper(clinvar, medgen, medgenhpo, omim).unwrap();
+            println!("The command has completed:{:?}", command);
+        }
+        Commands::NCBIANNOTATE {
+            ncbigeneid,
+            clinvar,
+            medgenomim,
+            medgenhpo,
+            omimsearch,
+        } => {
+            let command =
+                ncbiannotate(ncbigeneid, clinvar, medgenomim, medgenhpo, omimsearch).unwrap();
+            println!("The command has completed:{:?}", command);
+        }
+        Commands::ANNOTATOR {
+            pathncbimaxo,
+            medgenomim,
+            medgenhpo,
+            evidence,
+        } => {
+            let command = ontologyannotate(pathncbimaxo, medgenomim, medgenhpo, evidence).unwrap();
+            println!("The command has been completed:{:?}", command)
+        }
+        Commands::VCFCLINVARANNOTATE { vcffile, clinvar } => {
+            let command = clinvarvcf(vcffile, clinvar).unwrap();
+            println!(
+                "The command has finished and the annotated vcf file has been written:{:?}",
+                command
+            );
+        }
+        Commands::PHENOTYPELINKER {
+            genesdisease,
+            genesphenotype,
+            phenotypehpoa,
+            phenotypesgenes,
+        } => {
+            let command =
+                phenotypeall(genesdisease, genesphenotype, phenotypehpoa, phenotypesgenes).unwrap();
+            println!("The command has finished: {:?}", command);
+        }
+       Commands::Databases { databaseoption } => {
+                 let command = databasedownload(*databaseoption).unwrap();
+                 println!("The command has been finished and all the files have been downloaded:{}", command);
+       }
+    }
+}

src/args.rs

@@ -0,0 +1,28 @@
+use clap::{Parser, Subcommand};
+#[derive(Debug, Parser)]
+#[command(
+    name = "varlinker",
+    version = "1.0",
+    about = "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].
+      ************************************************"
+)]
+pub struct CommandParse {
+    /// subcommands for the specific actions
+    #[command(subcommand)]
+    pub command: Commands,
+}
+
+#[derive(Subcommand, Debug)]
+pub enum Commands {
+    /// annotate the specific coordinate
+    VARIANTLINKER {
+        /// variant VCF file
+        vcfile: String,
+    },
+}

src/main.rs

@@ -1,3 +1,29 @@
+mod args;
+mod store;
+mod variantlinker;
+use crate::args::CommandParse;
+use crate::args::Commands;
+use clap::Parser;
+
+/*
+ Authom GauravSablok
+ Instytut Chemii Bioorganicznej
+ Polskiej Akademii Nauk
+ ul. Noskowskiego 12/14 | 61-704, Poznań
+ Date: 2025-4-8
+*/
+
 fn main() {
-    println!("Hello, world!");
+    let argsparse = CommandParse::parse();
+    match &argsparse.command {
+        Commands::CUIGENERATE {
+            medgenhpo,
+            medgen_omim,
+            medgenmapping,
+            medgenpubmed,
+        } => {
+            let command = cuiparallel(medgenhpo, medgen_omim, medgenmapping, medgenpubmed).unwrap();
+            println!("The command has been completed:{:?}", command);
+        }
+    }
 }

src/store.rs

@@ -0,0 +1,16 @@
+#[derive(Debug, Clone, PartialOrd, PartialEq)]
+
+pub struct VCF {
+   pub  chrom: String,
+   pub  pos: usize,
+   pub  id: String,
+   pub refnuc: String,
+   pub altnuc: String,
+   pub qual: String,
+}
+
+#[derive(Debug, Clone, PartialOrd, PartialEq)]
+
+pub struct GENCODE {
+    pub
+}

src/variantlinker.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;
+
+
+/*
+ Authom GauravSablok
+ Instytut Chemii Bioorganicznej
+ Polskiej Akademii Nauk
+ ul. Noskowskiego 12/14 | 61-704, Poznań
+ Date: 2025-4-8
+*/
+
+pub async 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")
+        .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())
+}