feat: implementing ClinGen region dosage annotation (#282) (#319)

build.rs

@@ -24,6 +24,7 @@ fn main() -> Result<(), anyhow::Error> {
         "annonars/gnomad/vep_gnomad2.proto",
         "annonars/gnomad/vep_gnomad3.proto",
         "annonars/helixmtdb/base.proto",
+        "annonars/regions/clingen.proto",
     ]
     .iter()
     .map(|f| root.join(f))

protos/annonars/regions/clingen.proto

@@ -0,0 +1,25 @@
+// Genomic region annotation with ClinGen information.
+
+syntax = "proto3";
+
+package annonars.regions.clingen;
+
+import "annonars/genes/base.proto";
+
+// Dosage pathogenicity region annotation.
+message Region {
+    // ISCA ID
+    string isca_id = 1;
+    // ISCA Region Name
+    string isca_region_name = 2;
+    // Genomic locaion.
+    string genomic_location = 3;
+    // Haploinsufficiency score.
+    annonars.genes.base.ClingenDosageScore haploinsufficiency_score = 4;
+    // Triplosensitivity score.
+    annonars.genes.base.ClingenDosageScore triplosensitivity_score = 5;
+    // Haploinsufficiency Disease ID.
+    optional string haploinsufficiency_disease_id = 6;
+    // Haploinsufficiency Disease ID.
+    optional string triplosensitivity_disease_id = 7;
+}

src/genes/cli/data.rs

@@ -104,9 +104,9 @@ pub mod clingen_gene {
         }
     }
 
-    impl From<Score> for crate::pbs::genes::ClingenDosageScore {
+    impl From<Score> for crate::pbs::genes::base::ClingenDosageScore {
         fn from(val: Score) -> Self {
-            use crate::pbs::genes::ClingenDosageScore::*;
+            use crate::pbs::genes::base::ClingenDosageScore::*;
             match val {
                 Score::SufficientEvidence => SufficientEvidenceAvailable,
                 Score::SomeEvidence => SomeEvidenceAvailable,
@@ -1616,7 +1616,7 @@ pub mod shet {
 pub mod gtex {
     use serde::{Deserialize, Serialize};
 
-    use crate::pbs::genes::{GtexTissue, GtexTissueDetailed};
+    use crate::pbs::genes::base::{GtexTissue, GtexTissueDetailed};
 
     /// GTEx V8 tissue types.
     #[derive(Debug, Clone, Serialize, Deserialize)]

src/genes/cli/import.rs

@@ -371,7 +371,7 @@ fn load_domino(path: &str) -> Result<HashMap<String, domino::Record>, anyhow::Er
 }
 
 /// Convert from `data::*` records to protobuf records.
-fn convert_record(record: data::Record) -> pbs::genes::Record {
+fn convert_record(record: data::Record) -> pbs::genes::base::Record {
     let data::Record {
         acmg_sf,
         clingen,
@@ -402,7 +402,7 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             variants_to_report,
         } = acmg_sf;
 
-        pbs::genes::AcmgSecondaryFindingRecord {
+        pbs::genes::base::AcmgSecondaryFindingRecord {
             hgnc_id,
             ensembl_gene_id,
             ncbi_gene_id,
@@ -428,15 +428,15 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             triplosensitivity_disease_id,
         } = clingen;
 
-        pbs::genes::ClingenDosageRecord {
+        pbs::genes::base::ClingenDosageRecord {
             gene_symbol,
             ncbi_gene_id,
             genomic_location,
-            haploinsufficiency_score: Into::<pbs::genes::ClingenDosageScore>::into(
+            haploinsufficiency_score: Into::<pbs::genes::base::ClingenDosageScore>::into(
                 clingen_gene::Score::try_from(haploinsufficiency_score)
                     .expect("invalid haploinsufficiency score"),
             ) as i32,
-            triplosensitivity_score: Into::<pbs::genes::ClingenDosageScore>::into(
+            triplosensitivity_score: Into::<pbs::genes::base::ClingenDosageScore>::into(
                 clingen_gene::Score::try_from(triplosensitivity_score)
                     .expect("invalid triplosensitivity score"),
             ) as i32,
@@ -548,7 +548,7 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             zfin_zebrafish_phenotype_tag,
         } = dbnsfp;
 
-        pbs::genes::DbnsfpRecord {
+        pbs::genes::base::DbnsfpRecord {
             gene_name,
             ensembl_gene,
             chr,
@@ -680,7 +680,7 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             exac_oe_lof,
         } = gnomad_constraints;
 
-        pbs::genes::GnomadConstraintsRecord {
+        pbs::genes::base::GnomadConstraintsRecord {
             ensembl_gene_id,
             entrez_id,
             gene_symbol,
@@ -762,7 +762,7 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             mane_select,
         } = hgnc;
 
-        Some(pbs::genes::HgncRecord {
+        Some(pbs::genes::base::HgncRecord {
             hgnc_id,
             symbol,
             name,
@@ -796,7 +796,7 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             lsdb: lsdb
                 .map(|lsdb| {
                     lsdb.iter()
-                        .map(|lsdb| pbs::genes::HgncLsdb {
+                        .map(|lsdb| pbs::genes::base::HgncLsdb {
                             name: lsdb.name.clone(),
                             url: lsdb.url.clone(),
                         })
@@ -831,14 +831,14 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             summary,
             rif_entries,
         } = ncbi;
-        pbs::genes::NcbiRecord {
+        pbs::genes::base::NcbiRecord {
             gene_id,
             summary,
             rif_entries: rif_entries
                 .map(|rif_entries| {
                     rif_entries
                         .into_iter()
-                        .map(|rif_entry| pbs::genes::RifEntry {
+                        .map(|rif_entry| pbs::genes::base::RifEntry {
                             pmids: rif_entry.pmids.unwrap_or_default(),
                             text: rif_entry.text,
                         })
@@ -850,11 +850,11 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
 
     let omim = omim.map(|omim| {
         let omim::Record { hgnc_id, diseases } = omim;
-        pbs::genes::OmimRecord {
+        pbs::genes::base::OmimRecord {
             hgnc_id,
             omim_diseases: diseases
                 .into_iter()
-                .map(|disease| pbs::genes::OmimTerm {
+                .map(|disease| pbs::genes::base::OmimTerm {
                     omim_id: disease.omim_id,
                     label: disease.label,
                 })
@@ -864,11 +864,11 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
 
     let orpha = orpha.map(|orpha| {
         let orpha::Record { hgnc_id, diseases } = orpha;
-        pbs::genes::OrphaRecord {
+        pbs::genes::base::OrphaRecord {
             hgnc_id,
             orpha_diseases: diseases
                 .into_iter()
-                .map(|disease| pbs::genes::OrphaTerm {
+                .map(|disease| pbs::genes::base::OrphaTerm {
                     orpha_id: disease.orpha_id,
                     label: disease.label,
                 })
@@ -882,7 +882,7 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
             p_haplo,
             p_triplo,
         } = rcnv;
-        pbs::genes::RcnvRecord {
+        pbs::genes::base::RcnvRecord {
             hgnc_id,
             p_haplo,
             p_triplo,
@@ -891,7 +891,7 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
 
     let shet = shet.map(|shet| {
         let shet::Record { hgnc_id, s_het } = shet;
-        pbs::genes::ShetRecord { hgnc_id, s_het }
+        pbs::genes::base::ShetRecord { hgnc_id, s_het }
     });
 
     let gtex = gtex.map(|gtex| {
@@ -909,14 +909,14 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
                     tissue_detailed,
                     tpms,
                 } = record;
-                pbs::genes::GtexTissueRecord {
+                pbs::genes::base::GtexTissueRecord {
                     tissue: tissue as i32,
                     tissue_detailed: tissue_detailed as i32,
                     tpms,
                 }
             })
             .collect::<Vec<_>>();
-        pbs::genes::GtexRecord {
+        pbs::genes::base::GtexRecord {
             hgnc_id,
             ensembl_gene_id,
             ensembl_gene_version,
@@ -926,10 +926,10 @@ fn convert_record(record: data::Record) -> pbs::genes::Record {
 
     let domino = domino.map(|domino| {
         let domino::Record { gene_symbol, score } = domino;
-        pbs::genes::DominoRecord { gene_symbol, score }
+        pbs::genes::base::DominoRecord { gene_symbol, score }
     });
 
-    pbs::genes::Record {
+    pbs::genes::base::Record {
         acmg_sf,
         clingen,
         dbnsfp,

src/genes/cli/query.rs

@@ -63,7 +63,7 @@ pub fn open_rocksdb_from_args(
 fn print_record(
     out_writer: &mut Box<dyn std::io::Write>,
     output_format: common::cli::OutputFormat,
-    value: &genes::Record,
+    value: &genes::base::Record,
 ) -> Result<(), anyhow::Error> {
     match output_format {
         common::cli::OutputFormat::Jsonl => {
@@ -79,13 +79,13 @@ pub fn query_for_gene(
     hgnc_id: &str,
     db: &rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
     cf_data: &Arc<rocksdb::BoundColumnFamily>,
-) -> Result<Option<genes::Record>, anyhow::Error> {
+) -> Result<Option<genes::base::Record>, anyhow::Error> {
     let raw_value = db
         .get_cf(cf_data, hgnc_id.as_bytes())
         .map_err(|e| anyhow::anyhow!("error while querying for HGNC ID {}: {}", hgnc_id, e))?;
     raw_value
         .map(|raw_value| {
-            genes::Record::decode(&mut std::io::Cursor::new(&raw_value)).map_err(|e| {
+            genes::base::Record::decode(&mut std::io::Cursor::new(&raw_value)).map_err(|e| {
                 anyhow::anyhow!(
                     "error while decoding gene record for HGNC ID {}: {}",
                     hgnc_id,

src/lib.rs

@@ -19,6 +19,7 @@ pub mod gnomad_nuclear;
 pub mod gnomad_sv;
 pub mod helixmtdb;
 pub mod pbs;
+pub mod regions;
 pub mod server;
 pub mod tsv;
 

src/main.rs

@@ -1,6 +1,6 @@
 use annonars::{
     clinvar_genes, clinvar_minimal, clinvar_sv, common, cons, db_utils, dbsnp, freqs, functional,
-    genes, gnomad_mtdna, gnomad_nuclear, gnomad_sv, helixmtdb, server, tsv,
+    genes, gnomad_mtdna, gnomad_nuclear, gnomad_sv, helixmtdb, regions, server, tsv,
 };
 use anyhow::Error;
 use clap::{command, Args, Parser, Subcommand};
@@ -52,6 +52,8 @@ enum Commands {
     GnomadNuclear(GnomadNuclear),
     /// "gnomad-sv" sub commands
     GnomadSv(GnomadSv),
+    /// "regions" sub commands
+    Regions(Regions),
     /// "db-utils" sub commands
     DbUtils(DbUtils),
     /// "run-server" command.
@@ -279,6 +281,23 @@ enum GnomadSvCommands {
     Query(gnomad_sv::cli::query::Args),
 }
 
+/// Parsing of "regions" subcommands.
+#[derive(Debug, Args, Clone)]
+struct Regions {
+    /// The sub command to run
+    #[command(subcommand)]
+    command: RegionsCommands,
+}
+
+/// Enum supporting the parsing of "regions *" subcommands.
+#[derive(Debug, Subcommand, Clone)]
+enum RegionsCommands {
+    /// "import" sub command
+    Import(regions::cli::import::Args),
+    /// "query" sub command
+    Query(regions::cli::query::Args),
+}
+
 /// Parsing of "db-utils" subcommands.
 #[derive(Debug, Args, Clone)]
 struct DbUtils {
@@ -387,6 +406,10 @@ pub fn main() -> Result<(), anyhow::Error> {
                 GnomadSvCommands::Import(args) => gnomad_sv::cli::import::run(&cli.common, args)?,
                 GnomadSvCommands::Query(args) => gnomad_sv::cli::query::run(&cli.common, args)?,
             },
+            Commands::Regions(args) => match &args.command {
+                RegionsCommands::Import(args) => regions::cli::import::run(&cli.common, args)?,
+                RegionsCommands::Query(args) => regions::cli::query::run(&cli.common, args)?,
+            },
             Commands::DbUtils(args) => match &args.command {
                 DbUtilsCommands::Copy(args) => db_utils::cli::copy::run(&cli.common, args)?,
                 DbUtilsCommands::DumpMeta(args) => {

src/pbs/genes.rs

@@ -1,4 +1,7 @@
 //! Code generate for protobufs by `prost-build`.
 
-include!(concat!(env!("OUT_DIR"), "/annonars.genes.base.rs"));
-include!(concat!(env!("OUT_DIR"), "/annonars.genes.base.serde.rs"));
+/// Code generate for protobufs by `prost-build`.
+pub mod base {
+    include!(concat!(env!("OUT_DIR"), "/annonars.genes.base.rs"));
+    include!(concat!(env!("OUT_DIR"), "/annonars.genes.base.serde.rs"));
+}

src/pbs/mod.rs

@@ -7,3 +7,4 @@ pub mod functional;
 pub mod genes;
 pub mod gnomad;
 pub mod helixmtdb;
+pub mod regions;

src/pbs/regions.rs

@@ -0,0 +1,10 @@
+//! Code generate for protobufs by `prost-build`.
+
+/// Code generate for protobufs by `prost-build`.
+pub mod clingen {
+    include!(concat!(env!("OUT_DIR"), "/annonars.regions.clingen.rs"));
+    include!(concat!(
+        env!("OUT_DIR"),
+        "/annonars.regions.clingen.serde.rs"
+    ));
+}