feat: move clinvar database code (import etc.) to annonars (#86)

Cargo.toml

@@ -66,7 +66,7 @@ noodles-csi = "0.19"
 noodles-fasta = "0.23"
 noodles-bgzf = "0.22"
 noodles-tabix = "0.22"
-annonars = "0.7"
+annonars = "0.8"
 
 [build-dependencies]
 prost-build = "0.11.9"

src/annotate/seqvars/mod.rs

@@ -5,6 +5,7 @@ pub mod binning;
 pub mod csq;
 pub mod provider;
 
+use annonars::clinvar_minimal;
 use annonars::common::cli::is_canonical;
 use annonars::common::keys;
 use annonars::freqs::cli::import::reading::guess_assembly;
@@ -48,8 +49,6 @@ use thousands::Separable;
 use crate::annotate::seqvars::csq::{ConsequencePredictor, VcfVariant};
 use crate::annotate::seqvars::provider::MehariProvider;
 use crate::common::GenomeRelease;
-use crate::db::create::seqvar_clinvar::serialize::Record as ClinvarRecord;
-use crate::db::create::seqvar_clinvar::Pathogenicity;
 
 use crate::db::create::txs::data::TxSeqDatabase;
 use crate::ped::{PedigreeByName, Sex};
@@ -389,14 +388,19 @@ fn annotate_record_clinvar<T>(
 where
     T: ThreadMode,
 {
-    if let Some(clinvar_anno) = db.get_cf(cf, key)? {
-        let clinvar_record: ClinvarRecord = bincode::deserialize(&clinvar_anno)?;
+    if let Some(raw_value) = db.get_cf(cf, key)? {
+        let clinvar_record =
+            clinvar_minimal::pbs::Record::decode(&mut std::io::Cursor::new(&raw_value))?;
 
-        let ClinvarRecord {
+        let clinvar_minimal::pbs::Record {
             summary_clinvar_pathogenicity,
             vcv,
             ..
         } = clinvar_record;
+        let summary_clinvar_pathogenicity: Vec<_> = summary_clinvar_pathogenicity
+            .into_iter()
+            .map(|i: i32| -> clinvar_minimal::cli::reading::Pathogenicity { i.into() })
+            .collect();
 
         vcf_record.info_mut().insert(
             field::Key::from_str("clinvar_patho").unwrap(),
@@ -1079,8 +1083,9 @@ impl VarFishSeqvarTsvWriter {
             .unwrap_or_default()
             .map(|v| match v {
                 field::Value::String(value) => {
-                    Pathogenicity::from_str(value).unwrap_or(Pathogenicity::UncertainSignificance)
-                        >= Pathogenicity::LikelyPathogenic
+                    clinvar_minimal::cli::reading::Pathogenicity::from_str(value).unwrap_or(
+                        clinvar_minimal::cli::reading::Pathogenicity::UncertainSignificance,
+                    ) >= clinvar_minimal::cli::reading::Pathogenicity::LikelyPathogenic
                 }
                 _ => panic!("Unexpected value type for INFO/clinvar_patho"),
             })
@@ -1485,7 +1490,7 @@ fn run_with_writer(writer: &mut dyn AnnotatedVcfWriter, args: &Args) -> Result<(
             // Annotate with variant effect.
             if let Some(ann_fields) = predictor.predict(&VcfVariant {
                 chromosome: chrom,
-                position: pos as i32,
+                position: pos,
                 reference,
                 alternative,
             })? {

src/annotate/strucvars/mod.rs

@@ -186,7 +186,7 @@ pub mod vcf_header {
         };
 
         for sequence in &assembly_info.sequences {
-            if is_canonical(&sequence.name.as_ref()) {
+            if is_canonical(sequence.name.as_ref()) {
                 let mut contig = Map::<Contig>::try_from(vec![
                     (String::from("length"), format!("{}", sequence.length)),
                     (String::from("assembly"), assembly_name.clone()),
@@ -2402,7 +2402,7 @@ mod conv {
 
     /// Helper function to handle processing of "GT" field.
     fn process_gt(
-        entries: &mut Vec<GenotypeInfo>,
+        entries: &mut [GenotypeInfo],
         sample_no: usize,
         gt: &String,
         pedigree: &PedigreeByName,
@@ -2413,23 +2413,20 @@ mod conv {
         let sex = pedigree
             .individuals
             .get(&entries[sample_no].name)
-            .expect(&format!(
-                "sample must be in pedigree: {:?}",
-                &entries[sample_no].name
-            ))
+            .unwrap_or_else(|| panic!("sample must be in pedigree: {:?}", &entries[sample_no].name))
             .sex;
         let is_chr_x = tsv_record.chromosome.contains('X');
         let is_chr_y = tsv_record.chromosome.contains('Y');
         let has_ref = entries[sample_no]
             .gt
             .as_ref()
             .expect("just set")
-            .contains("0");
+            .contains('0');
         let has_alt = entries[sample_no]
             .gt
             .as_ref()
             .expect("just set")
-            .contains("1");
+            .contains('1');
 
         match (is_chr_x, is_chr_y, sex, has_ref, has_alt) {
             // autosomal chromosomes; treat "./." as missing, "./1" (and similar) as hom,
@@ -2458,9 +2455,7 @@ mod conv {
                 // count as hemi alt. on male chrY
                 tsv_record.num_hemi_alt += 1;
             }
-            (false, true, Sex::Female, _, _) => {
-                (/* do not count; variant in female */)
-            }
+            (false, true, Sex::Female, _, _) => {}
             (false, true, _, _, _) => (/* do not count; sex missing */),
             // conflicting chromosome
             (true, true, _, _, _) => panic!("cannot be both chrX and chrY"),
@@ -2475,14 +2470,14 @@ mod conv {
         let is_chr_x = tsv_record.chromosome.contains('X');
         let is_chr_y = tsv_record.chromosome.contains('Y');
         for entry in &tsv_record.genotype.entries {
-            let has_ref = entry.gt.as_ref().map(|s| s.contains("0")).unwrap_or(false);
-            let has_alt = entry.gt.as_ref().map(|s| s.contains("1")).unwrap_or(false);
+            let has_ref = entry.gt.as_ref().map(|s| s.contains('0')).unwrap_or(false);
+            let has_alt = entry.gt.as_ref().map(|s| s.contains('1')).unwrap_or(false);
             if !has_ref && !has_alt {
                 // This entry was previously removed, we can correct it now (if FORMAT/CN was set).
                 let sex = pedigree
                     .individuals
                     .get(&entry.name)
-                    .expect(&format!("sample must be in pedigree: {:?}", &entry.name))
+                    .unwrap_or_else(|| panic!("sample must be in pedigree: {:?}", &entry.name))
                     .sex;
                 let expected_cn = match (sex, is_chr_x, is_chr_y) {
                     (_, false, false) => Some(2),
@@ -3136,8 +3131,6 @@ mod test {
                 },
                 sex: if sample == "mother" {
                     Sex::Female
-                } else if sample == "father" {
-                    Sex::Male
                 } else {
                     Sex::Male
                 },

src/db/create/mod.rs

@@ -1,4 +1,3 @@
 //! Creation of mehari internal databases.
 
-pub mod seqvar_clinvar;
 pub mod txs;