pandora_lib_promethion/src/config.rs

#[derive(Debug, Clone)]
pub struct Config {
    pub pod_dir: String,
    pub result_dir: String,
    pub align: AlignConfig,
    pub reference: String,
    pub reference_name: String,
    pub savana_bin: String,
    pub savana_threads: u8,
    pub tumoral_name: String,
    pub normal_name: String,
    pub haplotagged_bam_tag_name: String,
    pub savana_output_dir: String,
    pub germline_phased_vcf: String,
    pub savana_passed_vcf: String,
    pub conda_sh: String,
    pub savana_force: bool,
    pub deepvariant_output_dir: String,
    pub severus_bin: String,
    pub severus_force: bool,
    pub severus_threads: u8,
    pub vntrs_bed: String,
    pub severus_pon: String,
    pub severus_output_dir: String,
    pub severus_solo_output_dir: String,
    pub longphase_bin: String,
    pub longphase_threads: u8,
    pub longphase_modcall_vcf: String,
    pub modkit_bin: String,
    pub modkit_summary_threads: u8,
    pub modkit_summary_file: String,
    pub longphase_modcall_threads: u8,
    pub deepvariant_threads: u8,
    pub deepvariant_bin_version: String,
    pub deepvariant_model_type: String,
    pub deepvariant_force: bool,
}

lazy_static! {
    static ref DEEPVARIANT_OUTPUT_NAME: &'static str = "{id}_{time}_DeepVariant.vcf.gz";
}

impl Default for Config {
    fn default() -> Self {
        Self {
            pod_dir: "/data/run_data".to_string(),
            align: Default::default(),

            // Reference genome
            reference: "/data/ref/hs1/chm13v2.0.fa".to_string(),
            reference_name: "hs1".to_string(),

            // File structure
            result_dir: "/data/longreads_basic_pipe".to_string(),
            tumoral_name: "diag".to_string(),
            normal_name: "mrd".to_string(),
            haplotagged_bam_tag_name: "hp".to_string(),

            germline_phased_vcf:
                "{result_dir}/{id}/diag/ClairS/clair3_normal_tumoral_germline_output_PS.vcf"
                    .to_string(),
            conda_sh: "/data/miniconda3/etc/profile.d/conda.sh".to_string(),

            // DeepVariant
            deepvariant_output_dir: "{result_dir}/{id}/{time}/DeepVariant".to_string(),
            deepvariant_threads: 155,
            deepvariant_bin_version: "1.8.0".to_string(),
            deepvariant_model_type: "ONT_R104".to_string(),
            deepvariant_force: false,

            // Savana
            savana_bin: "savana".to_string(),
            savana_threads: 150,
            savana_output_dir: "{result_dir}/{id}/diag/savana".to_string(),
            savana_passed_vcf: "{output_dir}/{id}_diag_savana_PASSED.vcf".to_string(),
            savana_force: true,

            // Severus
            severus_bin: "/data/tools/Severus/severus.py".to_string(),
            severus_threads: 32,
            vntrs_bed: "/data/ref/hs1/vntrs_chm13.bed".to_string(),
            severus_pon: "/data/ref/hs1/PoN_1000G_chm13.tsv.gz".to_string(),
            severus_output_dir: "{result_dir}/{id}/diag/severus".to_string(),
            severus_solo_output_dir: "{result_dir}/{id}/{time}/severus".to_string(),
            severus_force: true,

            // Longphase
            longphase_bin: "/data/tools/longphase_linux-x64".to_string(),
            longphase_threads: 150,
            longphase_modcall_threads: 8, // ! out of memory
            longphase_modcall_vcf:
                "{result_dir}/{id}/{time}/5mC_5hmC/{id}_{time}_5mC_5hmC_modcall.vcf.gz".to_string(),

            // modkit
            modkit_bin: "modkit".to_string(),
            modkit_summary_threads: 50,
            modkit_summary_file: "{result_dir}/{id}/{time}/{id}_{time}_5mC_5hmC_summary.txt"
                .to_string(),
        }
    }
}

#[derive(Debug, Clone)]
pub struct AlignConfig {
    pub dorado_bin: String,
    pub dorado_basecall_arg: String,
    pub ref_fa: String,
    pub ref_mmi: String,
    pub samtools_view_threads: u16,
    pub samtools_sort_threads: u16,
}

impl Default for AlignConfig {
    fn default() -> Self {
        Self {
            dorado_bin: "/data/tools/dorado-0.9.0-linux-x64/bin/dorado".to_string(),
            dorado_basecall_arg: "-x 'cuda:0,1,2,3' sup,5mC_5hmC".to_string(), // since v0.8.0 need
            // to specify cuda devices (exclude the T1000)
            ref_fa: "/data/ref/hs1/chm13v2.0.fa".to_string(),
            ref_mmi: "/data/ref/chm13v2.0.mmi".to_string(),
            samtools_view_threads: 20,
            samtools_sort_threads: 50,
        }
    }
}

impl Config {
    pub fn tumoral_dir(&self, id: &str) -> String {
        format!("{}/{}/{}", self.result_dir, id, self.tumoral_name)
    }

    pub fn normal_dir(&self, id: &str) -> String {
        format!("{}/{}/{}", self.result_dir, id, self.normal_name)
    }

    pub fn solo_dir(&self, id: &str, time: &str) -> String {
        format!("{}/{}/{}", self.result_dir, id, time)
    }

    pub fn solo_bam(&self, id: &str, time: &str) -> String {
        format!(
            "{}/{}_{}_{}.bam",
            self.solo_dir(id, time),
            id,
            time,
            self.reference_name,
        )
    }

    pub fn tumoral_bam(&self, id: &str) -> String {
        format!(
            "{}/{}_{}_{}.bam",
            self.tumoral_dir(id),
            id,
            self.tumoral_name,
            self.reference_name,
        )
    }

    pub fn normal_bam(&self, id: &str) -> String {
        format!(
            "{}/{}_{}_{}.bam",
            self.normal_dir(id),
            id,
            self.normal_name,
            self.reference_name,
        )
    }

    pub fn tumoral_haplotagged_bam(&self, id: &str) -> String {
        format!(
            "{}/{}_{}_{}_{}.bam",
            self.tumoral_dir(id),
            id,
            self.tumoral_name,
            self.reference_name,
            self.haplotagged_bam_tag_name
        )
    }

    pub fn normal_haplotagged_bam(&self, id: &str) -> String {
        format!(
            "{}/{}_{}_{}_{}.bam",
            self.normal_dir(id),
            id,
            self.normal_name,
            self.reference_name,
            self.haplotagged_bam_tag_name
        )
    }

    pub fn germline_phased_vcf(&self, id: &str) -> String {
        self.germline_phased_vcf
            .replace("{result_dir}", &self.result_dir)
            .replace("{id}", id)
    }

    // DeepVariant
    pub fn deepvariant_output_dir(&self, id: &str, time: &str) -> String {
        self.deepvariant_output_dir
            .replace("{result_dir}", &self.result_dir)
            .replace("{id}", id)
            .replace("{time}", time)
    }

    pub fn deepvariant_output_vcf(&self, id: &str, time: &str) -> String {
        format!(
            "{}/{}",
            self.deepvariant_output_dir(id, time),
            *DEEPVARIANT_OUTPUT_NAME
        )
        .replace("{id}", id)
        .replace("{time}", time)
    }

    // Savana
    pub fn savana_output_dir(&self, id: &str) -> String {
        self.savana_output_dir
            .replace("{result_dir}", &self.result_dir)
            .replace("{id}", id)
    }

    pub fn savana_output_vcf(&self, id: &str) -> String {
        let output_dir = self.savana_output_dir(id);

        format!("{output_dir}/{id}_diag_hs1_hp.classified.somatic.vcf")
    }

    pub fn savana_passed_vcf(&self, id: &str) -> String {
        self.savana_passed_vcf
            .replace("{output_dir}", &self.savana_output_dir(id))
            .replace("{id}", id)
    }

    // Severus
    pub fn severus_output_dir(&self, id: &str) -> String {
        self.severus_output_dir
            .replace("{result_dir}", &self.result_dir)
            .replace("{id}", id)
    }

    pub fn severus_output_vcf(&self, id: &str) -> String {
        let output_dir = self.severus_output_dir(id);
        format!("{output_dir}/somatic_SVs/severus_somatic.vcf")
    }

    pub fn severus_passed_vcf(&self, id: &str) -> String {
        format!(
            "{}/{}_diag_severus_PASSED.vcf.gz",
            &self.severus_output_dir(id),
            id
        )
    }

    // Severus solo
    pub fn severus_solo_output_dir(&self, id: &str, time: &str) -> String {
        self.severus_solo_output_dir
            .replace("{result_dir}", &self.result_dir)
            .replace("{id}", id)
            .replace("{time}", time)
    }

    pub fn severus_solo_output_vcf(&self, id: &str, time: &str) -> String {
        let output_dir = self.severus_solo_output_dir(id, time);
        format!("{output_dir}/all_SVs/severus_all.vcf")
    }

    pub fn severus_solo_passed_vcf(&self, id: &str, time: &str) -> String {
        format!(
            "{}/{}_{}_severus-solo_PASSED.vcf.gz",
            &self.severus_solo_output_dir(id, time),
            id,
            time
        )
    }

    pub fn constit_vcf(&self, id: &str) -> String {
        format!("{}/{}_variants_constit.vcf.gz", self.tumoral_dir(id), id)
    }

    pub fn constit_phased_vcf(&self, id: &str) -> String {
        format!("{}/{}_variants_constit_PS.vcf.gz", self.tumoral_dir(id), id)
    }

    pub fn modkit_summary_file(&self, id: &str, time: &str) -> String {
        self.modkit_summary_file
            .replace("{result_dir}", &self.result_dir)
            .replace("{id}", id)
            .replace("{time}", time)
    }

    pub fn longphase_modcall_vcf(&self, id: &str, time: &str) -> String {
        self.longphase_modcall_vcf
            .replace("{result_dir}", &self.result_dir)
            .replace("{id}", id)
            .replace("{time}", time)
    }
}