pandora_lib_promethion/src/lib.rs

use std::sync::{Arc, Mutex};

pub mod commands;
pub mod config;
pub mod modkit;
pub mod callers;
pub mod runners;
pub mod collection;
pub mod functions;
pub mod helpers;

#[macro_use]
extern crate lazy_static;

// Define DOCKER_ID lock for handling Docker kill when ctrlc is pressed
lazy_static! {
    static ref DOCKER_ID: Arc<Mutex<Vec<String>>> = Arc::new(Mutex::new(Vec::new()));
}

#[cfg(test)]
mod tests {
    use std::{fs, path::Path};

    use callers::{nanomonsv::nanomonsv_create_pon, savana::Savana, severus::Severus};
    use collection::{InitSomatic, Version};
    use commands::{longphase::{LongphaseHap, LongphaseConfig}, modkit::{bed_methyl, ModkitConfig}};
    use functions::assembler::{Assembler, AssemblerConfig};
    use log::info;
    // use pandora_lib_assembler::assembler::AssembleConfig;
    use rayon::prelude::*;
    use runners::Run;

    use self::{callers::deep_variant::DeepVariantConfig, collection::pod5::{FlowCellCase, Pod5Collection}, commands::dorado, config::Config};
    use super::*;
    use crate::{callers::{clairs::{ClairS, ClairSConfig}, deep_variant::DeepVariant, nanomonsv::{NanomonSV, NanomonSVConfig, NanomonSVSolo}}, collection::{bam::{self, BamType}, run_tasks, variants::VariantsCollection, vcf::VcfCollection, Collections, CollectionsConfig}, commands::{bcftools::{bcftools_keep_pass, BcftoolsConfig}, dorado::Dorado}};

    // export RUST_LOG="debug"
    fn init() {
        let _ = env_logger::Builder::from_env(env_logger::Env::default().default_filter_or("info"))
            .is_test(true)
            .try_init();
    }

    #[test]
    fn it_works() {
        let bam_path = "/data/longreads_basic_pipe/PARACHINI/diag/PARACHINI_diag_hs1.bam";
        modkit::modkit(bam_path);
    }

    #[test]
    fn run_dorado() -> anyhow::Result<()> {
        let case = FlowCellCase { 
            id: "CONSIGNY".to_string(), time_point: "mrd".to_string(), barcode: "07".to_string(), pod_dir: "/data/run_data/20240326-CL/CONSIGNY-MRD-NB07_RICCO-DIAG-NB08/20240326_1355_1E_PAU78333_bc25da25/pod5_pass/barcode07".into() };
        dorado::Dorado::init(case, Config::default())?.run_pipe()
    }

    #[test]
    fn pod5() -> anyhow::Result<()> {
        let _ = env_logger::Builder::from_env(env_logger::Env::default().default_filter_or("info"))
            .build();

        let _ = Pod5Collection::new(
            "/data/run_data",
            "/data/flow_cells.tsv",
            "/data/longreads_basic_pipe",
        )?;
        // let runs = Runs::import_dir("/home/prom/store/banana-pool/run_data", "/data/flow_cells.tsv")?;
        Ok(())
    }

    #[test_log::test]
    fn bam() -> anyhow::Result<()> {
        let bam_collection = bam::load_bam_collection("/data/longreads_basic_pipe");

        bam_collection
            .bams
            .iter()
            .filter(|b| matches!(b.bam_type, BamType::Panel(_)))
            .for_each(|b| println!("{b:#?}"));

        let u = bam_collection.get("PARACHINI", "mrd");
        println!("{u:#?}");

        Ok(())
    }

    #[test_log::test]
    fn vcf() -> anyhow::Result<()> {
        let mut vcf_collection = VcfCollection::new("/data/longreads_basic_pipe");
        vcf_collection.sort_by_id();
        vcf_collection
            .vcfs
            .iter()
            .for_each(|v| v.println().unwrap());

        Ok(())
    }

    // pod5 view -I /data/run_data/20240903-CL/ARMEM-DG-N02_ASSJU-DG-N03/20240903_1428_1B_PAW47629_fc24c3cf/pod5/PAW47629_fc24c3cf_77b07847_0.pod5 | head -5000 | awk '{if(NR==1){print "target,"$0}else{print "subset_1.pod5,"$0}}' > /tmp/subset_ids.csv
    // pod5 subset /data/run_data/20240903-CL/ARMEM-DG-N02_ASSJU-DG-N03/20240903_1428_1B_PAW47629_fc24c3cf/pod5/PAW47629_fc24c3cf_77b07847_0.pod5 --csv /tmp/subset_ids.csv -o /data/test_suite/pod5/muxed/
    #[test_log::test]
    fn mux() -> anyhow::Result<()> {
        let result_dir = "/data/test_suite/results".to_string();
        let cases = vec![
            FlowCellCase { id: "test_02".to_string(), time_point: "diag".to_string(), barcode: "02".to_string(), pod_dir: "/data/test_suite/pod5/muxed".into() },
            FlowCellCase { id: "test_03".to_string(), time_point: "diag".to_string(), barcode: "03".to_string(), pod_dir: "/data/test_suite/pod5/muxed".into() },
        ];

        cases.iter().for_each(|c| {
            let dir = format!("{result_dir}/{}", c.id);
            if Path::new(&dir).exists() {
                fs::remove_dir_all(dir).unwrap();
            }
        });
        let config = Config { result_dir, ..Default::default()  };
        Dorado::from_mux(cases, config)
    }

    #[test_log::test]
    fn deep_variant() -> anyhow::Result<()> {
        // let config = DeepVariantConfig {
        //     result_dir: "/data/test".to_string(),
        //     ..DeepVariantConfig::default()
        // };
        // DeepVariant::new("test_a", "diag", "/data/test_data/subset.bam", config).run()
        let config = DeepVariantConfig {
            result_dir: "/data/test".to_string(),
            ..DeepVariantConfig::default()
        };
        DeepVariant::new("LEVASSEUR", "mrd", "/data/longreads_basic_pipe/LEVASSEUR/mrd/LEVASSEUR_mrd_hs1.bam", config).run()
    }

    #[test_log::test]
    fn clairs() -> anyhow::Result<()> {
        let config = ClairSConfig {
            result_dir: "/data/test".to_string(),
            ..ClairSConfig::default()
        };
        ClairS::new("test_a", "/data/test_data/subset.bam", "/data/test_data/subset_mrd.bam", config).run()
    }

    #[test_log::test]
    fn nanomonsv() -> anyhow::Result<()> {
        // let config = NanomonSVConfig {
        //     result_dir: "/data/test".to_string(),
        //     ..NanomonSVConfig::default()
        // };
        // NanomonSV::new("test_a", "/data/test_data/subset.bam", "/data/test_data/subset_mrd.bam", config).run()

        let bam = |id:&str, time_point: &str| format!("/data/longreads_basic_pipe/{id}/{time_point}/{id}_{time_point}_hs1.bam");
        NanomonSV::new("CAZIER", &bam("CAZIER", "diag"), &bam("CAZIER", "mrd"), NanomonSVConfig::default()).run()
    }

    #[test]
    fn nanomonsv_solo() -> anyhow::Result<()> {
        init();
        let id = "BRETON";
        let time_point = "diag";
        NanomonSVSolo::new(id, &format!("/data/longreads_basic_pipe/{id}/{time_point}/{id}_{time_point}_hs1.bam"), time_point, NanomonSVConfig::default()).run()?;
        // let time_point = "mrd";
        // for id in ["MERY", "CAMARA", "FRANIATTE", "FERATI", "IQBAL", "COLLE", "JOLIVET", "BAFFREAU", "MANCUS", "BELARBI", "BENGUIRAT", "HENAUX", "MEDDAH"] {
        //
        // NanomonSVSolo::new(id, &format!("/data/longreads_basic_pipe/{id}/{time_point}/{id}_{time_point}_hs1.bam"), time_point, NanomonSVConfig::default()).run()?;
        // }
        Ok(())
    }

    // cargo test run -- --nocapture; ~/run_scripts/notify_finish.sh &
    #[test]
    fn todo_all() -> anyhow::Result<()> {
        init();
        // let config = CollectionsConfig::default();
        let config = CollectionsConfig { pod_dir: "/data/store".to_string(), ..Default::default() };
        info!("Runing todo with config: {:#?}", config);
        let mut collections = Collections::new(config)?;
        collections.todo()?;
        collections.tasks.iter().for_each(|t| println!("{t}"));
        println!("{}", collections.tasks.len());
        Ok(())
    }

    #[test]
    fn todo_agg() -> anyhow::Result<()> {
        init();
        let config = CollectionsConfig::default();
        info!("Runing todo with config: {:#?}", config);
        let collections = Collections::new(config)?;
        let agg_tasks = collections.todo_variants_agg()?;
        println!("{:#?}", agg_tasks);
        println!("{}", agg_tasks.len());
        Ok(())
    }

    #[test]
    fn run_agg() -> anyhow::Result<()> {
        init();
        let config = CollectionsConfig {
            id_black_list: vec!["MANCUSO".to_string(),"HAMROUNE".to_string()],
            ..Default::default()
        };
        info!("Runing todo with config: {:#?}", config);
        let mut collections = Collections::new(config)?;
        collections.tasks = collections.todo_variants_agg()?;
        collections.run()?;
        
        Ok(())
    }

    // export RUST_LOG="debug"
    #[test_log::test]
    fn run_t() -> anyhow::Result<()> {
        // let config = CollectionsConfig::default();
        let config = CollectionsConfig { pod_dir: "/data/store".to_string(), ..Default::default() };

        run_tasks(config)
    }

    #[test_log::test]
    fn somatic() -> anyhow::Result<()> {
        let variants_collection = VariantsCollection::new("/data/longreads_basic_pipe")?;
        variants_collection.data.iter().for_each(|v| println!("{}\t{}", v.id, v.path.display()));
        Ok(())
    }

    #[test_log::test]
    fn bcftools_pass() {
        let config = BcftoolsConfig::default();
        let id = "RICCO";
        let i = format!("/data/longreads_basic_pipe/{id}/diag/nanomonsv/{id}_diag.nanomonsv.result.vcf");
        let o = format!("/data/longreads_basic_pipe/{id}/diag/nanomonsv/{id}_diag_nanomonsv_PASSED.vcf.gz");
        bcftools_keep_pass(&i, &o, config).unwrap();
    }

    #[test_log::test]
    fn bam_ok() -> anyhow::Result<()> {
        let  collections = Collections::new(
            CollectionsConfig::default()
        )?;
        let mut res: Vec<_> = collections.bam.by_id_completed(15.0, 10.0).iter().map(|b| {
            (b.id.to_string(), b.time_point.to_string(), b.path.to_str().unwrap().to_string())
        }).collect();
        res.sort_by_key(|b| b.1.clone());
        res.sort_by_key(|b| b.0.clone());
        
        res.iter().for_each(|(id, tp, path)| println!("{id}\t{tp}\t{path}"));
        Ok(())
    }

    #[test_log::test]
    fn todo_assembler() -> anyhow::Result<()> {
        let collections = Collections::new(
            CollectionsConfig::default()
        )?;
        collections.todo_assembler()?;
        Ok(())
    }

    #[test]
    fn sv_pon() -> anyhow::Result<()> {
        init();
        nanomonsv_create_pon(NanomonSVConfig::default(), "/data/ref/hs1/nanomonsv_pon.vcf.gz")
    }

    #[test]
    fn todo_mod() -> anyhow::Result<()> {
        init();
        let collections = Collections::new(
            CollectionsConfig::default()
        )?;
        collections.todo_mod_pileup();
        Ok(())
    }

    #[test]
    fn todo_deepv() -> anyhow::Result<()> {
        init();
        let collections = Collections::new(
            CollectionsConfig::default()
        )?;
        let tasks = collections.todo_deepvariants();
        tasks.iter().for_each(|t| info!("{t}"));
        info!("n tasks {}", tasks.len());
        Ok(())
    }

    #[test]
    fn todo_clairs() -> anyhow::Result<()> {
        init();
        let collections = Collections::new(
            CollectionsConfig::default()
        )?;
        collections.todo_clairs().iter().for_each(|t| info!("{t}"));
        Ok(())
    }

    #[test]
    fn run_assemblers() -> anyhow::Result<()> {
        Assembler::new("CAMEL".to_string(), "diag".to_string(), AssemblerConfig::default()).run()
    }

    // #[test]
    // fn run_dmr_par() -> anyhow::Result<()> {
    //     init();
    //     let collections = Collections::new(
    //         CollectionsConfig::default()
    //     )?;
    //     let tasks = collections.todo_dmr_c_diag_mrd();
    //     tasks.iter().for_each(|t| info!("{t}"));
    //     let len = tasks.len();
    //     // let pool = ThreadPoolBuilder::new().num_threads(10).build().unwrap();
    //     // pool.install(|| {
    //     //     tasks.par_iter().enumerate().for_each(|(i, t)| {
    //     //         let config = ModkitConfig {threads: 2,  ..Default::default() };
    //     //         if let collection::CollectionsTasks::DMRCDiagMrd { id, .. } = t { let _ = dmr_c_mrd_diag(id, &config); }
    //     //         println!("⚡ {i}/{len}");
    //     //     });
    //     // });
    //     Ok(())
    // }

    #[test]
    fn run_mod_par() -> anyhow::Result<()> {
        init();
        let collections = Collections::new(
            CollectionsConfig::default()
        )?;
        let tasks = collections.todo_mod_pileup();
        let len = tasks.len();

        tasks.par_iter().enumerate().for_each(|(i, t)| {
            let config = ModkitConfig {threads: 2,  ..Default::default() };
            if let collection::CollectionsTasks::ModPileup { bam, .. } = t { let _ = bed_methyl(bam.to_owned(), &config); }
            println!("⚡ {i}/{len}");
        });
        Ok(())
    }

    #[test]
    fn run_severus() -> anyhow::Result<()> {
        init();
        Severus::initialize("ACHITE", Config::default())?.run()
    }

    #[test]
    fn run_savana() -> anyhow::Result<()> {
        init();
        Savana::initialize("BECERRA", Config::default())?.run()
    }

    #[test]
    fn check_versions() -> anyhow::Result<()> {
        init();
        let config = Config::default();
        let v = Savana::version(&config)?;
        info!("Savanna version {v}");
        let v = Severus::version(&config)?;
        info!("Severus version {v}");
        Ok(())
    }

    #[test]
    fn run_multi_deepvariant() -> anyhow::Result<()> {
        init();
        let mut collections = Collections::new(
            CollectionsConfig::default()
        )?;
        collections.run_deepvariant()
    }

    #[test]
    fn run_clairs() -> anyhow::Result<()> {
        init();
        let collections = Collections::new(
            CollectionsConfig::default()
        )?;
        collections.run_clairs()
    }

    #[test]
    fn run_longphase() -> anyhow::Result<()> {
        init();
        let id = "CUNY";
        let diag_bam = format!("/data/longreads_basic_pipe/{id}/diag/{id}_diag_hs1.bam");
        let vcf = format!("/data/longreads_basic_pipe/{id}/diag/ClairS/clair3_normal_tumoral_germline_output.vcf.gz");
        let mrd_bam = format!("/data/longreads_basic_pipe/{id}/mrd/{id}_mrd_hs1.bam");

        LongphaseHap::new(id, &diag_bam, &vcf, LongphaseConfig::default()).run()?;
        LongphaseHap::new(id, &mrd_bam, &vcf, LongphaseConfig::default()).run()
    }
}