feat: adding support for gnomAD-SV (#291) (#297)

.gitattributes

@@ -1 +1,3 @@
 tests/**/*.db/** filter=lfs diff=lfs merge=lfs -text
+tests/**/rocksdb/** filter=lfs diff=lfs merge=lfs -text
+src/gnomad_sv/cli/snapshots/*.snap filter=lfs diff=lfs merge=lfs -text

build.rs

@@ -7,9 +7,13 @@ fn main() {
     println!("cargo:rerun-if-changed=src/proto/annonars/cons/v1/base.proto");
     println!("cargo:rerun-if-changed=src/proto/annonars/dbsnp/v1/base.proto");
     println!("cargo:rerun-if-changed=src/proto/annonars/gene/v1/base.proto");
-    println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/mtdna.proto");
+    println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/exac_cnv.proto");
     println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/gnomad2.proto");
     println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/gnomad3.proto");
+    println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/gnomad_cnv4.proto");
+    println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/gnomad_sv2.proto");
+    println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/gnomad_sv4.proto");
+    println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/mtdna.proto");
     println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/vep_common.proto");
     println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/vep_gnomad2.proto");
     println!("cargo:rerun-if-changed=src/proto/annonars/gnomad/v1/vep_gnomad3.proto");
@@ -43,9 +47,13 @@ fn main() {
                 "annonars/cons/v1/base.proto",
                 "annonars/dbsnp/v1/base.proto",
                 "annonars/gene/v1/base.proto",
-                "annonars/gnomad/v1/mtdna.proto",
+                "annonars/gnomad/v1/exac_cnv.proto",
                 "annonars/gnomad/v1/gnomad2.proto",
                 "annonars/gnomad/v1/gnomad3.proto",
+                "annonars/gnomad/v1/gnomad_cnv4.proto",
+                "annonars/gnomad/v1/gnomad_sv2.proto",
+                "annonars/gnomad/v1/gnomad_sv4.proto",
+                "annonars/gnomad/v1/mtdna.proto",
                 "annonars/gnomad/v1/vep_common.proto",
                 "annonars/gnomad/v1/vep_gnomad2.proto",
                 "annonars/gnomad/v1/vep_gnomad3.proto",

src/clinvar_sv/cli/query.rs

@@ -11,7 +11,7 @@ use crate::common::{self, cli::extract_chrom, spdi};
 #[derive(clap::Args, Debug, Clone, Default)]
 #[group(required = true, multiple = false)]
 pub struct ArgsQuery {
-    /// Specify range to query for.
+    /// Specify accession to query for.
     #[arg(long, group = "query")]
     pub accession: Option<String>,
     /// Query for all variants.
@@ -22,9 +22,9 @@ pub struct ArgsQuery {
     pub range: Option<spdi::Range>,
 }
 
-/// Command line arguments for `tsv query` sub command.
+/// Command line arguments for `clinvar-sv query` sub command.
 #[derive(clap::Parser, Debug, Clone)]
-#[command(about = "query gnomAD-mtDNA data stored in RocksDB", long_about = None)]
+#[command(about = "query ClinVar SV data stored in RocksDB", long_about = None)]
 pub struct Args {
     /// Path to RocksDB directory with data.
     #[arg(long)]

src/common/cli.rs

@@ -9,17 +9,21 @@ pub struct Args {
 }
 
 /// Output format to write.
-#[derive(Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Debug, clap::ValueEnum, strum::Display)]
+#[derive(
+    Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Debug, Default, clap::ValueEnum, strum::Display,
+)]
 #[strum(serialize_all = "lowercase")]
 pub enum OutputFormat {
     /// JSONL format.
+    #[default]
     Jsonl,
 }
 
 /// Local genome release for command line arguments.
 #[derive(
     Copy,
     Clone,
+    Default,
     PartialEq,
     Eq,
     PartialOrd,
@@ -34,6 +38,7 @@ pub enum OutputFormat {
 pub enum GenomeRelease {
     /// GRCh37 genome release.
     #[strum(serialize = "grch37")]
+    #[default]
     Grch37,
     /// GRCh38 genome release.
     #[strum(serialize = "grch38")]

src/common/spdi.rs

@@ -17,7 +17,7 @@ use std::{fmt::Display, str::FromStr};
 ///
 /// # Example
 ///
-/// ```
+/// ```ignore
 /// use std::str::FromStr;
 /// use annonars::common::spdi::Var;
 ///
@@ -84,7 +84,7 @@ impl Display for Var {
 ///
 /// # Example
 ///
-/// ```
+/// ```ignore
 /// use std::str::FromStr;
 /// use annonars::common::spdi::Pos;
 ///
@@ -133,7 +133,7 @@ impl Display for Pos {
 ///
 /// # Example
 ///
-/// ```
+/// ```ignore
 /// use std::str::FromStr;
 /// use annonars::common::spdi::Range;
 ///

src/gnomad_nuclear/cli/import.rs

@@ -14,23 +14,47 @@ use crate::{
 };
 
 /// Select the type of gnomAD data to import.
-#[derive(strum::Display, clap::ValueEnum, Debug, Clone, Copy, PartialEq, Eq, Hash)]
+#[derive(strum::Display, clap::ValueEnum, Debug, Clone, Copy, Default, PartialEq, Eq, Hash)]
 pub enum GnomadKind {
-    /// gnomAD exomes
+    /// gnomAD exomes / CNV (or: ExAC CNVs)
     #[strum(serialize = "exomes")]
+    #[default]
     Exomes,
-    /// gnomAD genomes
+    /// gnomAD genomes / SVs
     #[strum(serialize = "genomes")]
     Genomes,
 }
 
 /// Select the genomAD version (v2/v3; important for the field names).
 #[derive(strum::Display, clap::ValueEnum, Debug, Clone, Copy, PartialEq, Eq, Hash)]
 pub enum GnomadVersion {
+    /// Version 1.x (ExAC)
+    One,
     /// Version 2.x
     Two,
     /// Version 3.x
     Three,
+    /// Version 4.x
+    Four,
+}
+
+impl FromStr for GnomadVersion {
+    type Err = anyhow::Error;
+
+    /// Parse out the gnomAD version from the leading number.
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        Ok(if s.starts_with("1.") {
+            GnomadVersion::One
+        } else if s.starts_with("2.") {
+            GnomadVersion::Two
+        } else if s.starts_with("3.") {
+            GnomadVersion::Three
+        } else if s.starts_with("4.") {
+            GnomadVersion::Four
+        } else {
+            anyhow::bail!("invalid GnomadVersion: {}", s)
+        })
+    }
 }
 
 /// Command line arguments for `gnomad_nuclear import` sub command.
@@ -199,6 +223,7 @@ fn process_window(
                         )?
                         .encode_to_vec()
                     }
+                    _ => anyhow::bail!("gnomAD version must be either 2 or 3"),
                 };
                 db.put_cf(&cf_gnomad, &key_buf, &record_buf)?;
             }
@@ -210,13 +235,10 @@ fn process_window(
 
 /// Implementation of `gnomad_nuclear import` sub command.
 pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error> {
-    let gnomad_version = if args.gnomad_version.starts_with("2.") {
-        GnomadVersion::Two
-    } else if args.gnomad_version.starts_with("3.") {
-        GnomadVersion::Three
-    } else {
-        anyhow::bail!("gnomAD version must be either 2 or 3")
-    };
+    let gnomad_version: GnomadVersion = args.gnomad_version.parse()?;
+    if !matches!(gnomad_version, GnomadVersion::Two | GnomadVersion::Three) {
+        anyhow::bail!("gnomAD version must be either 2 or 3");
+    }
 
     // Put defaults for fields to serialize into args.
     let args = match gnomad_version {
@@ -242,6 +264,7 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
                 .transpose()?,
             ..args.clone()
         },
+        _ => anyhow::bail!("gnomAD version must be either 2 or 3"),
     };
 
     tracing::info!("Starting 'gnomad-nuclear import' command");

src/gnomad_pbs/exac_cnv.rs

@@ -0,0 +1,3 @@
+//! Code for working with the ExAC r1 CNV data.
+
+include!(concat!(env!("OUT_DIR"), "/annonars.gnomad.v1.exac_cnv.rs"));

src/gnomad_pbs/gnomad2.rs

@@ -349,7 +349,7 @@ impl Record {
                     by_sex: Some(AlleleCountsBySex {
                         overall: Self::extract_allele_counts(record, &prefix, "")?,
                         xx: Self::extract_allele_counts(record, &prefix, "_female")?,
-                        xy: Self::extract_allele_counts(record, &prefix, "male")?,
+                        xy: Self::extract_allele_counts(record, &prefix, "_male")?,
                     }),
                     raw: Self::extract_allele_counts(record, &prefix, "_raw")?,
                     popmax: common::noodles::get_string(record, &format!("{}_popmax", cohort)).ok(),

src/gnomad_pbs/gnomad_cnv4.rs

@@ -0,0 +1,6 @@
+//! Code for working with the gnomAD CNV v4.0 data.
+
+include!(concat!(
+    env!("OUT_DIR"),
+    "/annonars.gnomad.v1.gnomad_cnv4.rs"
+));

src/gnomad_pbs/gnomad_sv2.rs

@@ -0,0 +1,6 @@
+//! Code for working with the gnomAD SV v2.1 data.
+
+include!(concat!(
+    env!("OUT_DIR"),
+    "/annonars.gnomad.v1.gnomad_sv2.rs"
+));

src/gnomad_pbs/gnomad_sv4.rs

@@ -0,0 +1,6 @@
+//! Code for working with the gnomAD SV v4.0 data.
+
+include!(concat!(
+    env!("OUT_DIR"),
+    "/annonars.gnomad.v1.gnomad_sv4.rs"
+));

src/gnomad_pbs/mod.rs

@@ -1,7 +1,11 @@
 //! Protocolbuffers related code for gnomAD mtDNA and nuclear.
 
+pub mod exac_cnv;
 pub mod gnomad2;
 pub mod gnomad3;
+pub mod gnomad_cnv4;
+pub mod gnomad_sv2;
+pub mod gnomad_sv4;
 pub mod mtdna;
 pub mod vep_common;
 pub mod vep_gnomad2;