adding gnomad v4 example files

src/strucvars/txt_to_bin/cli.rs

@@ -72,8 +72,12 @@ pub enum InputType {
     StrucvarExacCnv,
     /// Convert Thousand Genomes to binary.
     StrucvarG1k,
-    /// Convert gnomAD SV to binary.
-    StrucvarGnomadSv,
+    /// Convert gnomAD SV v2 to binary.
+    StrucvarGnomadSv2,
+    /// Convert gnomAD CNV v4 to binary.
+    StrucvarGnomadCnv4,
+    /// Convert gnomAD SV v4 to binary.
+    StrucvarGnomadSv4,
     /// Convert masked region to binary.
     MaskedRegion,
     /// Convert cross-link to binary.
@@ -100,7 +104,7 @@ pub struct Args {
 
 /// Main entry point for the `strucvars txt-to-bin` command.
 pub fn run(common_args: &crate::common::Args, args: &Args) -> Result<(), anyhow::Error> {
-    tracing::info!("Starting `db to-bin`");
+    tracing::info!("Starting `strucvars txt-to-bin`");
     tracing::info!("  common_args = {:?}", &common_args);
     tracing::info!("  args = {:?}", &args);
 
@@ -135,11 +139,21 @@ pub fn run(common_args: &crate::common::Args, args: &Args) -> Result<(), anyhow:
         InputType::StrucvarG1k => {
             vardbs::convert_to_bin(&args.path_input, &args.path_output, InputFileType::G1k)?
         }
-        InputType::StrucvarGnomadSv => vardbs::convert_to_bin(
+        InputType::StrucvarGnomadSv2 => vardbs::convert_to_bin(
             &args.path_input,
             &args.path_output,
             InputFileType::GnomadSv2,
         )?,
+        InputType::StrucvarGnomadCnv4 => vardbs::convert_to_bin(
+            &args.path_input,
+            &args.path_output,
+            InputFileType::GnomadCnv4,
+        )?,
+        InputType::StrucvarGnomadSv4 => vardbs::convert_to_bin(
+            &args.path_input,
+            &args.path_output,
+            InputFileType::GnomadSv4,
+        )?,
         InputType::MaskedRegion => masked::convert_to_bin(&args.path_input, &args.path_output)?,
         InputType::Xlink => xlink::convert_to_bin(&args.path_input, &args.path_output)?,
     }
@@ -303,14 +317,14 @@ mod test {
     }
 
     #[test]
-    fn run_strucvar_gnomad_smoke() -> Result<(), anyhow::Error> {
+    fn run_strucvar_gnomad_sv2_smoke() -> Result<(), anyhow::Error> {
         let tmp_dir = temp_testdir::TempDir::default();
         let common_args = common::Args {
             verbose: Verbosity::new(0, 0),
         };
         let args = Args {
             assembly: None,
-            input_type: InputType::StrucvarGnomadSv,
+            input_type: InputType::StrucvarGnomadSv2,
             path_input: String::from(
                 "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/gnomad_sv.bed.gz",
             ),
@@ -322,6 +336,46 @@ mod test {
         Ok(())
     }
 
+    #[test]
+    fn run_strucvar_gnomad_cnv4_smoke() -> Result<(), anyhow::Error> {
+        let tmp_dir = temp_testdir::TempDir::default();
+        let common_args = common::Args {
+            verbose: Verbosity::new(0, 0),
+        };
+        let args = Args {
+            assembly: None,
+            input_type: InputType::StrucvarGnomadCnv4,
+            path_input: String::from(
+                "tests/db/to-bin/varfish-db-downloader/vardbs/grch38/strucvar/gnomad-cnv.bed.gz",
+            ),
+            path_output: tmp_dir.join("gnomad-cnv.bin"),
+        };
+
+        super::run(&common_args, &args)?;
+
+        Ok(())
+    }
+
+    #[test]
+    fn run_strucvar_gnomad_sv4_smoke() -> Result<(), anyhow::Error> {
+        let tmp_dir = temp_testdir::TempDir::default();
+        let common_args = common::Args {
+            verbose: Verbosity::new(0, 0),
+        };
+        let args = Args {
+            assembly: None,
+            input_type: InputType::StrucvarGnomadSv4,
+            path_input: String::from(
+                "tests/db/to-bin/varfish-db-downloader/vardbs/grch38/strucvar/gnomad-sv.bed.gz",
+            ),
+            path_output: tmp_dir.join("gnomad-sv.bin"),
+        };
+
+        super::run(&common_args, &args)?;
+
+        Ok(())
+    }
+
     #[test]
     fn run_masked_region_smoke() -> Result<(), anyhow::Error> {
         let tmp_dir = temp_testdir::TempDir::default();

src/strucvars/txt_to_bin/vardbs/input.rs

@@ -113,10 +113,26 @@ pub struct GnomadSv4Record {
     pub end: i32,
     /// The structural vairant type
     pub svtype: String,
-    /// Number of homozygous alternative carriers
-    pub n_homalt: u32,
-    /// Number of heterozygous carriers
-    pub n_het: u32,
+    /// Number of male homozygous reference allele carriers.
+    pub male_n_homref: u32,
+    /// Number of male heterozygous alternate allele carriers.
+    pub male_n_het: u32,
+    /// Number of male homozygous alternate allele carriers.
+    pub male_n_homalt: u32,
+    /// Number of male hemizygous alternate allele carriers.
+    pub male_n_hemiref: u32,
+    /// Number of male hemizygous reference allele carriers.
+    pub male_n_hemialt: u32,
+    /// Number of female homozygous reference allele carriers.
+    pub female_n_homref: u32,
+    /// Number of female heterozygous alternate allele carriers.
+    pub female_n_het: u32,
+    /// Number of female homozygous alternate allele carriers.
+    pub female_n_homalt: u32,
+    /// Number of samples at this site (CNV only).
+    pub cnv_n_total: u32,
+    /// Number of samples with a CNV at this site (CNV only).
+    pub cnv_n_var: u32,
 }
 
 /// gnomAD CNV v$ database record as read from TSV file.
@@ -130,10 +146,10 @@ pub struct GnomadCnv4Record {
     pub end: i32,
     /// The structural vairant type
     pub svtype: String,
-    /// Number of homozygous alternative carriers
-    pub n_homalt: u32,
-    /// Number of heterozygous carriers
-    pub n_het: u32,
+    /// Number of samples at this site (passing QC).
+    pub n_total: u32,
+    /// Number of samples with a CNV at this site (passing QC).
+    pub n_var: u32,
 }
 
 /// Common type to convert input data to.
@@ -317,15 +333,52 @@ impl TryInto<Option<InputRecord>> for GnomadCnv4Record {
     type Error = &'static str;
 
     fn try_into(self) -> Result<Option<InputRecord>, Self::Error> {
-        todo!()
+        Ok(Some(InputRecord {
+            chromosome: self.chromosome.clone(),
+            chromosome2: self.chromosome,
+            begin: self.begin,
+            end: self.end,
+            sv_type: match self.svtype.as_str() {
+                "DEL" => SvType::Del,
+                "DUP" => SvType::Dup,
+                _ => {
+                    error!("sv_type = {}", &self.svtype);
+                    return Err("unknown SV type");
+                }
+            },
+            count: self.n_var,
+        }))
     }
 }
 
 impl TryInto<Option<InputRecord>> for GnomadSv4Record {
     type Error = &'static str;
 
     fn try_into(self) -> Result<Option<InputRecord>, Self::Error> {
-        todo!()
+        Ok(Some(InputRecord {
+            chromosome: self.chromosome.clone(),
+            chromosome2: self.chromosome,
+            begin: self.begin,
+            end: self.end,
+            sv_type: match self.svtype.as_str() {
+                "BND" => SvType::Bnd,
+                "CNV" => SvType::Cnv,
+                "DEL" => SvType::Del,
+                "DUP" => SvType::Dup,
+                "INS" => SvType::Ins,
+                "INV" => SvType::Inv,
+                _ => {
+                    error!("sv_type = {}", &self.svtype);
+                    return Err("unknown SV type");
+                }
+            },
+            count: self.male_n_het
+                + self.male_n_homalt
+                + self.male_n_hemialt
+                + self.female_n_het
+                + self.female_n_homalt
+                + self.cnv_n_var,
+        }))
     }
 }
 

src/strucvars/txt_to_bin/vardbs/mod.rs

@@ -145,39 +145,39 @@ mod test {
     use super::InputFileType;
 
     #[rstest::rstest]
-    #[case(
+    #[case::dbvar(
         InputFileType::Dbvar,
         "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/dbvar.bed.gz"
     )]
-    #[case(
+    #[case::dgv(
         InputFileType::Dgv,
         "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/dgv.bed.gz"
     )]
-    #[case(
+    #[case::dgv_gs(
         InputFileType::DgvGs,
         "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/dgv_gs.bed.gz"
     )]
-    #[case(
+    #[case::exac(
         InputFileType::Exac,
         "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/exac.bed.gz"
     )]
-    #[case(
+    #[case::g1k(
         InputFileType::G1k,
         "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/g1k.bed.gz"
     )]
-    #[case(
+    #[case::gnomad_sv2(
         InputFileType::GnomadSv2,
         "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/gnomad_sv.bed.gz"
     )]
-    // #[case(
-    //     InputFileType::GnomadCnv4,
-    //     "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/gnomad_cnv4.bed.gz",
-    // )]
-    // #[case(
-    //     InputFileType::GnomadSv4,
-    //     "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/gnomad_sv4.bed.gz",
-    // )]
-    #[case(
+    #[case::gnomad_cnv4(
+        InputFileType::GnomadCnv4,
+        "tests/db/to-bin/varfish-db-downloader/vardbs/grch38/strucvar/gnomad-cnv.bed.gz"
+    )]
+    #[case::gnomad_sv4(
+        InputFileType::GnomadSv4,
+        "tests/db/to-bin/varfish-db-downloader/vardbs/grch38/strucvar/gnomad-sv.bed.gz"
+    )]
+    #[case::inhouse_db(
         InputFileType::InhouseDb,
         "tests/db/to-bin/varfish-db-downloader/vardbs/grch37/strucvar/inhouse.tsv"
     )]