pandora_lib_variants/src/annotations/echtvar.rs

use std::{process::{Command, Stdio}, io::{BufReader, BufRead}};

use anyhow::{Context, Ok, Result};
use log::{info, warn};

use super::{cosmic::Cosmic, gnomad::GnomAD};

// /data/tools/echtvar anno -e /data/ref/hs1/CosmicCodingMuts.echtvar.zip -e /data/ref/hs1/gnomAD_4-2022_10-gnomad.echtvar.zip BENGUIRAT_diag_clairs_PASSED.vcf.gz test.bcf
pub fn run_echtvar(in_path: &str, out_path: &str) -> Result<()> {
    let bin_dir = "/data/tools";

    let annot_sources: Vec<&str> = vec![
        "/data/ref/hs1/CosmicCodingMuts.echtvar.zip",
        "/data/ref/hs1/gnomAD_4-2022_10-gnomad.echtvar.zip",
    ]
    .iter()
    .flat_map(|e| vec!["-e", e])
    .collect();

    // info!("Running echtvar anno for {}", in_path);
    let mut cmd = Command::new(format!("{}/echtvar", bin_dir))
        .arg("anno")
        .args(annot_sources)
        .arg(in_path)
        .arg(out_path)
        .stderr(Stdio::piped())
        .spawn()
        .context("echtvar anno failed to start")?;

    let stderr = cmd.stderr.take().unwrap();
    let reader = BufReader::new(stderr);
    reader
        .lines()
        .filter_map(|line| line.ok())
        .filter(|line| line.find("error").is_some())
        .for_each(|line| warn!("{}", line));

    cmd.wait()?;
    Ok(())
}

pub fn parse_echtvar_val(s: &str) -> Result<(Option<Cosmic>, Option<GnomAD>)> {
    let mi: Vec<_> = s.match_indices(r";gnomad").collect();
    let (i, _) = mi[0];
    let str_cosmic = &s[..i];
    let str_gnomad = &s[i + 1..];

    let cosmic = if str_cosmic.contains("MISSING") {
        None
    } else {
        Some(str_cosmic.parse::<Cosmic>()?)
    };

    let gnomad = if str_gnomad.contains("-1") {
        None
    } else {
        Some(str_gnomad.parse::<GnomAD>()?)
    };
    Ok((cosmic, gnomad))
}