pandora_lib_promethion/src/commands/longphase.rs

use crate::{
    collection::{Initialize, InitializeSolo},
    commands::bcftools::{bcftools_compress, bcftools_index},
    config::Config,
    helpers::path_prefix,
    runners::{run_wait, CommandRun, Run},
};
use anyhow::Context;
use duct::cmd;
use std::{
    fs,
    path::{Path, PathBuf},
};
use tracing::info;

use super::{
    bcftools::{bcftools_keep_pass, BcftoolsConfig},
    modkit::ModkitSummary,
};

#[derive(Debug, Clone)]
pub struct LongphaseConfig {
    pub bin: String,
    pub result_dir: String,
    pub reference: String,
    pub threads: u8,
    pub force: bool,
}

impl Default for LongphaseConfig {
    fn default() -> Self {
        Self {
            bin: "/data/tools/longphase_linux-x64".to_string(),
            reference: "/data/ref/hs1/chm13v2.0.fa".to_string(),
            result_dir: "/data/longreads_basic_pipe".to_string(),
            threads: 150,
            force: true,
        }
    }
}

#[derive(Debug)]
pub struct LongphaseHap {
    pub id: String,
    pub vcf: String,
    pub bam: PathBuf,
    pub bam_hp: PathBuf,
    pub config: LongphaseConfig,
    pub log_dir: String,
}

impl LongphaseHap {
    pub fn new(id: &str, bam: &str, phased_vcf: &str, config: LongphaseConfig) -> Self {
        let log_dir = format!("{}/{}/log/longphase", config.result_dir, id);

        let bam = Path::new(bam);
        // TODO change that use config.haplotagged_bam_tag_name
        let new_fn = format!("{}_HP", bam.file_stem().unwrap().to_str().unwrap());
        let bam_hp = bam.with_file_name(new_fn);

        Self {
            id: id.to_string(),
            bam: bam.to_path_buf(),
            config,
            log_dir,
            vcf: phased_vcf.to_string(),
            bam_hp: bam_hp.to_path_buf(),
        }
    }

    pub fn run(&mut self) -> anyhow::Result<()> {
        if self.config.force && self.bam_hp.exists() {
            fs::remove_file(&self.bam_hp)?;
        }

        if !Path::new(&self.log_dir).exists() {
            fs::create_dir_all(&self.log_dir).expect("Failed to create output directory");
        }

        // Run command if output VCF doesn't exist
        if !self.bam_hp.exists() {
            let args = [
                "haplotag",
                "-s",
                &self.vcf,
                "-b",
                self.bam.to_str().unwrap(),
                "-r",
                &self.config.reference,
                "-t",
                &self.config.threads.to_string(),
                "--tagSupplementary",
                "-o",
                self.bam_hp.to_str().unwrap(),
            ];
            let mut cmd_run = CommandRun::new(&self.config.bin, &args);
            let report = run_wait(&mut cmd_run).context(format!(
                "Error while running `{} {}`",
                self.config.bin,
                args.join(" ")
            ))?;

            let log_file = format!("{}/longphase_", self.log_dir);
            report
                .save_to_file(&log_file)
                .context(format!("Error while writing logs into {log_file}"))?;

            let _ = cmd!(
                "samtools",
                "index",
                "-@",
                &self.config.threads.to_string(),
                &format!("{}.bam", self.bam_hp.to_str().unwrap())
            )
            .run()?;
        } else {
            info!("Longphase output vcf already exists");
        }

        Ok(())
    }
}

// /data/tools/longphase_linux-x64 phase -s ClairS/clair3_normal_tumoral_germline_output.vcf.gz -b CUNY_diag_hs1_hp.bam -r /data/ref/hs1/chm13v2.0.fa -t 155 --ont -o ClairS/clair3_normal_tumoral_germline_output_PS
#[derive(Debug)]
pub struct LongphasePhase {
    pub id: String,
    pub vcf: String,
    pub out_prefix: String,
    pub bam: String,
    pub config: Config,
    pub log_dir: String,
    pub modcall_vcf: String,
}

impl Initialize for LongphasePhase {
    fn initialize(id: &str, config: crate::config::Config) -> anyhow::Result<Self> {
        let log_dir = format!("{}/{}/log/longphase_phase", config.result_dir, id);
        if !Path::new(&log_dir).exists() {
            fs::create_dir_all(&log_dir)
                .context(format!("Failed  to create {log_dir} directory"))?;
        }
        let vcf = config.constit_vcf(id);
        let bam = config.tumoral_bam(id);
        let out_prefix = path_prefix(&config.constit_phased_vcf(id))?;
        let modcall_vcf = config.longphase_modcall_vcf(id, "diag");

        Ok(LongphasePhase {
            id: id.to_string(),
            config,
            log_dir,
            vcf,
            out_prefix,
            bam,
            modcall_vcf,
        })
    }
}

impl Run for LongphasePhase {
    fn run(&mut self) -> anyhow::Result<()> {
        info!("Running longphase phase for: {}", self.vcf);
        info!("Saving longphase phase results in: {}", self.out_prefix);

        let final_vcf = self.config.constit_phased_vcf(&self.id);
        if !Path::new(&final_vcf).exists() {
            let args = [
                "phase",
                "-s",
                &self.vcf,
                "-b",
                &self.bam,
                "-r",
                &self.config.reference,
                "--mod-file",
                &self.modcall_vcf,
                "-t",
                &self.config.longphase_threads.to_string(),
                "--ont",
                "-o",
                &self.out_prefix,
            ];
            let mut cmd_run = CommandRun::new(&self.config.longphase_bin, &args);
            let report = run_wait(&mut cmd_run).context(format!(
                "Error while running `{} {}`",
                self.config.longphase_bin,
                args.join(" ")
            ))?;

            let log_file = format!("{}/longphase_phase_", self.log_dir);
            report
                .save_to_file(&log_file)
                .context(format!("Error while writing logs into {log_file}"))?;

            bcftools_compress(
                &format!("{}.vcf", self.out_prefix),
                &final_vcf,
                &BcftoolsConfig::default(),
            )?;
            bcftools_index(&final_vcf, &BcftoolsConfig::default())?;
            fs::remove_file(format!("{}.vcf", self.out_prefix))?;
        }
        Ok(())
    }
}

#[derive(Debug)]
pub struct LongphaseModcallSolo {
    pub id: String,
    pub time: String,
    pub bam: String,
    pub prefix: String,
    pub reference: String,
    pub threads: u8,
    pub log_dir: String,
    pub mod_threshold: f64,
    pub config: Config,
}

impl InitializeSolo for LongphaseModcallSolo {
    fn initialize(id: &str, time: &str, config: Config) -> anyhow::Result<Self> {
        let id = id.to_string();
        let time = time.to_string();

        let log_dir = format!("{}/{}/log/longphase_modcall_solo", config.result_dir, &id);
        if !Path::new(&log_dir).exists() {
            fs::create_dir_all(&log_dir)
                .context(format!("Failed  to create {log_dir} directory"))?;
        }

        let bam = config.solo_bam(&id, &time);
        if !Path::new(&bam).exists() {
            anyhow::bail!("Bam files doesn't exists: {bam}")
        }

        let mut modkit_summary = ModkitSummary::initialize(&id, &time, config.clone())?;
        modkit_summary.load()?;
        let mod_threshold = modkit_summary
            .result
            .ok_or_else(|| anyhow::anyhow!("Error no ModkitSummary for {id} {time}"))?
            .pass_threshold;

        let out_vcf = config.longphase_modcall_vcf(&id, &time);
        let out_dir = Path::new(&out_vcf)
            .parent()
            .ok_or_else(|| anyhow::anyhow!("Can't get dir of {out_vcf}"))?;
        fs::create_dir_all(out_dir)?;
        let prefix = path_prefix(&out_vcf)?;

        Ok(Self {
            id,
            time,
            bam,
            reference: config.reference.to_string(),
            threads: config.longphase_modcall_threads,
            config,
            log_dir,
            mod_threshold,
            prefix,
        })
    }
}

impl Run for LongphaseModcallSolo {
    fn run(&mut self) -> anyhow::Result<()> {
        let args = [
            "modcall",
            "-b",
            &self.bam,
            "-t",
            &self.threads.to_string(),
            "-r",
            &self.reference,
            "-m",
            &self.mod_threshold.to_string(),
            "-o",
            &self.prefix,
        ];
        let mut cmd_run = CommandRun::new(&self.config.longphase_bin, &args);
        run_wait(&mut cmd_run)
            .context(format!(
                "Error while running `longphase modcall {}`",
                args.join(" ")
            ))?
            .save_to_file(&format!("{}/longphase_modcall_", self.log_dir))
            .context(format!(
                "Error while writing logs into {}/longphase_modcall",
                self.log_dir
            ))?;

        let vcf = format!("{}.vcf", self.prefix);
        bcftools_keep_pass(
            &vcf,
            &format!("{}.vcf.gz", self.prefix),
            BcftoolsConfig::default(),
        )
        .context(format!(
            "Can't run BCFtools PASS for LongphaseModcallSolo: {} {}",
            self.id, self.time
        ))?
        .save_to_file(&format!("{}/longphase_modcall_pass_", self.log_dir))
        .context(format!(
            "Error while writing logs into {}/longphase_modcall_pass",
            self.log_dir
        ))?;
        fs::remove_file(&vcf).context(format!("Can't remove file: {vcf}"))?;
        Ok(())
    }
}