use std::path::Path;
use std::sync::Arc;
use crate::common::contig::ContigManager;
use crate::db::keys;
use annonars::common::contig::ContigDict;
use anyhow::Error;
use prost::Message;
use rocksdb::{DBWithThreadMode, MultiThreaded};
#[derive(Debug)]
pub struct CaddAnnotator {
db: DBWithThreadMode<MultiThreaded>,
contig_manager: Arc<ContigManager>,
contig_dict: ContigDict,
}
#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
pub struct CaddResult {
pub raw_score: f32,
pub phred: f32,
}
impl CaddAnnotator {
pub fn new(
db: DBWithThreadMode<MultiThreaded>,
contig_manager: Arc<ContigManager>,
contig_dict: ContigDict,
) -> Self {
Self {
db,
contig_manager,
contig_dict,
}
}
pub(crate) fn from_path(
path: impl AsRef<Path>,
contig_manager: Arc<ContigManager>,
) -> anyhow::Result<Self> {
tracing::info!("Opening CADD database at {}", path.as_ref().display());
let options = rocksdb::Options::default();
let db_cadd = rocksdb::DB::open_cf_for_read_only(&options, &path, ["meta", "cadd"], false)?;
let contig_dict = annonars::seqvars::read_contig_dict(&db_cadd)?;
Ok(Self::new(db_cadd, contig_manager, contig_dict))
}
pub fn annotate_record_cadd(&self, key: &[u8]) -> Result<Option<CaddResult>, Error> {
if let Some(raw_value) = self
.db
.get_cf(self.db.cf_handle("cadd").as_ref().unwrap(), key)?
{
let record = annonars::pbs::seqvars::base::CaddRecord::decode(&raw_value[..])?;
Ok(Some(CaddResult {
raw_score: record.raw_score,
phred: record.phred,
}))
} else {
Ok(None)
}
}
pub fn annotate(&self, vcf_var: &keys::Var) -> anyhow::Result<Option<CaddResult>> {
let chrom_std = self
.contig_manager
.get_primary_name(&vcf_var.chrom)
.cloned()
.unwrap_or_else(|| vcf_var.chrom.clone());
if let Some(chrom_id) = self.contig_dict.id_of(&chrom_std) {
let annonars_var: annonars::common::keys::Var = vcf_var.into();
let key = annonars_var.encode_with_id(chrom_id);
self.annotate_record_cadd(&key)
} else {
Ok(None)
}
}
}
#[cfg(test)]
mod tests {
use super::*;
use temp_testdir::TempDir;
#[test]
fn test_cadd_import_and_annotate() -> Result<(), anyhow::Error> {
let temp = TempDir::default();
let fai_path = temp.join("ref.fa.fai");
let input_path = temp.join("cadd_test.tsv");
let output_path = temp.join("cadd_db");
std::fs::write(&fai_path, "chr1\t248956422\t6\t60\t61\n")?;
std::fs::write(
&input_path,
"#Chrom\tPos\tRef\tAlt\tRawScore\tPHRED\n\
1\t10000\tC\tT\t1.5\t15.0\n\
chr1\t10005\tA\tG\t2.3\t20.5\n",
)?;
let common_args = annonars::common::cli::Args {
verbose: Default::default(),
};
let import_args = annonars::cadd::cli::import::Args {
path_in_tsv: input_path.to_string_lossy().into(),
path_reference_fai: fai_path.to_string_lossy().into(),
assembly: "GRCh38".to_string(),
path_out_rocksdb: output_path.to_string_lossy().into(),
cf_name: "cadd".to_string(),
path_wal_dir: None,
};
annonars::cadd::cli::import::run(&common_args, &import_args)?;
let contig_manager = Arc::new(ContigManager::new("GRCh38"));
let annotator = CaddAnnotator::from_path(&output_path, contig_manager)?;
let var1 = keys::Var {
chrom: "chr1".to_string(),
pos: 10000,
reference: "C".to_string(),
alternative: "T".to_string(),
};
let res1 = annotator.annotate(&var1)?.unwrap();
assert_eq!(res1.raw_score, 1.5);
assert_eq!(res1.phred, 15.0);
let var2 = keys::Var {
chrom: "1".to_string(),
pos: 10005,
reference: "A".to_string(),
alternative: "G".to_string(),
};
let res2 = annotator.annotate(&var2)?.unwrap();
assert_eq!(res2.raw_score, 2.3);
assert_eq!(res2.phred, 20.5);
let var3 = keys::Var {
chrom: "1".to_string(),
pos: 99999,
reference: "C".to_string(),
alternative: "G".to_string(),
};
assert!(annotator.annotate(&var3)?.is_none());
Ok(())
}
}