wip

varfish-org · Dec 28, 2023 · 276f217 · 276f217
1 parent 5517687
commit 276f217
Show file tree

Hide file tree

Showing 26 changed files with 255 additions and 22 deletions.
diff --git a/src/gnomad_nuclear/cli/import.rs b/src/gnomad_nuclear/cli/import.rs
@@ -25,9 +25,9 @@ pub enum GnomadKind {
     Genomes,
 }
 
-impl Into<crate::pbs::gnomad::gnomad4::RecordType> for GnomadKind {
-    fn into(self) -> crate::pbs::gnomad::gnomad4::RecordType {
-        match self {
+impl From<GnomadKind> for crate::pbs::gnomad::gnomad4::RecordType {
+    fn from(val: GnomadKind) -> Self {
+        match val {
             GnomadKind::Exomes => crate::pbs::gnomad::gnomad4::RecordType::Exomes,
             GnomadKind::Genomes => crate::pbs::gnomad::gnomad4::RecordType::Genomes,
         }
@@ -414,7 +414,7 @@ pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error>
     for (header_field, val) in more_header_values {
         db.put_cf(
             &cf_meta,
-            &format!("gnomad-{}", header_field.replace("_", "-")),
+            &format!("gnomad-{}", header_field.replace('_', "-")),
             &val,
         )?;
     }
@@ -504,7 +504,7 @@ mod test {
     }
 
     #[test]
-    fn smoke_test_import_gnomad_exomes_grch38() -> Result<(), anyhow::Error> {
+    fn smoke_test_import_gnomad_exomes_grch38_v2_1() -> Result<(), anyhow::Error> {
         let tmp_dir = TempDir::default();
         let common = common::cli::Args {
             verbose: Verbosity::new(1, 0),
@@ -529,7 +529,7 @@ mod test {
     }
 
     #[test]
-    fn smoke_test_import_gnomad_genomes_grch38() -> Result<(), anyhow::Error> {
+    fn smoke_test_import_gnomad_genomes_grch38_v3_1() -> Result<(), anyhow::Error> {
         let tmp_dir = TempDir::default();
         let common = common::cli::Args {
             verbose: Verbosity::new(1, 0),
@@ -552,4 +552,54 @@ mod test {
 
         run(&common, &args)
     }
+
+    #[test]
+    fn smoke_test_import_gnomad_exomes_grch38_v4_0() -> Result<(), anyhow::Error> {
+        let tmp_dir = TempDir::default();
+        let common = common::cli::Args {
+            verbose: Verbosity::new(1, 0),
+        };
+        let args = Args {
+            genome_release: common::cli::GenomeRelease::Grch38,
+            path_in_vcf: vec![String::from(
+                "tests/gnomad-nuclear/example-exomes-grch38/v4.0/gnomad-exomes.vcf.bgz",
+            )],
+            path_out_rocksdb: format!("{}", tmp_dir.join("out-rocksdb").display()),
+            cf_name: String::from("gnomad_nuclear_data"),
+            path_wal_dir: None,
+            tbi_window_size: 1_000_000,
+            import_fields_json: Some(serde_json::to_string(
+                &gnomad3::DetailsOptions::with_all_enabled(),
+            )?),
+            gnomad_kind: GnomadKind::Exomes,
+            gnomad_version: String::from("4.0"),
+        };
+
+        run(&common, &args)
+    }
+
+    #[test]
+    fn smoke_test_import_gnomad_genomes_grch38_v4_0() -> Result<(), anyhow::Error> {
+        let tmp_dir = TempDir::default();
+        let common = common::cli::Args {
+            verbose: Verbosity::new(1, 0),
+        };
+        let args = Args {
+            genome_release: common::cli::GenomeRelease::Grch38,
+            path_in_vcf: vec![String::from(
+                "tests/gnomad-nuclear/example-genomes-grch38/v4.0/gnomad-genomes.vcf.bgz",
+            )],
+            path_out_rocksdb: format!("{}", tmp_dir.join("out-rocksdb").display()),
+            cf_name: String::from("gnomad_nuclear_data"),
+            path_wal_dir: None,
+            tbi_window_size: 1_000_000,
+            import_fields_json: Some(serde_json::to_string(
+                &gnomad3::DetailsOptions::with_all_enabled(),
+            )?),
+            gnomad_kind: GnomadKind::Genomes,
+            gnomad_version: String::from("4.0"),
+        };
+
+        run(&common, &args)
+    }
 }
diff --git a/src/gnomad_nuclear/cli/query.rs b/src/gnomad_nuclear/cli/query.rs
@@ -291,8 +291,9 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    // #[case("exomes", "grch38", "4.0")]
-    fn smoke_query_exomes_all(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_all(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
@@ -316,15 +317,24 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    fn smoke_query_exomes_var_single(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_var_single(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
     ) -> Result<(), anyhow::Error> {
         crate::common::set_snapshot_suffix!("{}-{}-{}", kind, genome_release, version);
         let (common, args, _temp) = build_args(
             ArgsQuery {
-                variant: Some(spdi::Var::from_str("GRCh37:1:55516888:G:GA")?),
+                variant: Some(spdi::Var::from_str(&format!(
+                    "{}:1:55516888:G:GA",
+                    if genome_release == "grch37" {
+                        "GRCh37"
+                    } else {
+                        "GRCh38"
+                    }
+                ))?),
                 ..Default::default()
             },
             kind,
@@ -340,15 +350,24 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    fn smoke_query_exomes_pos_single(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_pos_single(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
     ) -> Result<(), anyhow::Error> {
         crate::common::set_snapshot_suffix!("{}-{}-{}", kind, genome_release, version);
         let (common, args, _temp) = build_args(
             ArgsQuery {
-                position: Some(spdi::Pos::from_str("GRCh37:1:55516888")?),
+                position: Some(spdi::Pos::from_str(&format!(
+                    "{}:1:55516888",
+                    if genome_release == "grch37" {
+                        "GRCh37"
+                    } else {
+                        "GRCh38"
+                    }
+                ))?),
                 ..Default::default()
             },
             kind,
@@ -364,15 +383,24 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    fn smoke_query_exomes_range_find_all_chr1(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_range_find_all_chr1(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
     ) -> Result<(), anyhow::Error> {
         crate::common::set_snapshot_suffix!("{}-{}-{}", kind, genome_release, version);
         let (common, args, _temp) = build_args(
             ArgsQuery {
-                range: Some(spdi::Range::from_str("GRCh37:1:1:249250621")?),
+                range: Some(spdi::Range::from_str(&format!(
+                    "{}:1:1:249250621",
+                    if genome_release == "grch37" {
+                        "GRCh37"
+                    } else {
+                        "GRCh38"
+                    }
+                ))?),
                 ..Default::default()
             },
             kind,
@@ -388,15 +416,24 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    fn smoke_query_exomes_range_find_first(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_range_find_first(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
     ) -> Result<(), anyhow::Error> {
         crate::common::set_snapshot_suffix!("{}-{}-{}", kind, genome_release, version);
         let (common, args, _temp) = build_args(
             ArgsQuery {
-                range: Some(spdi::Range::from_str("GRCh37:1:55505599:55505599")?),
+                range: Some(spdi::Range::from_str(&format!(
+                    "{}:1:55505599:55505599",
+                    if genome_release == "grch37" {
+                        "GRCh37"
+                    } else {
+                        "GRCh38"
+                    }
+                ))?),
                 ..Default::default()
             },
             kind,
@@ -412,15 +449,24 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    fn smoke_query_exomes_range_find_second(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_range_find_second(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
     ) -> Result<(), anyhow::Error> {
         crate::common::set_snapshot_suffix!("{}-{}-{}", kind, genome_release, version);
         let (common, args, _temp) = build_args(
             ArgsQuery {
-                range: Some(spdi::Range::from_str("GRCh37:1:55505615:55505615")?),
+                range: Some(spdi::Range::from_str(&format!(
+                    "{}:1:55505615:55505615",
+                    if genome_release == "grch37" {
+                        "GRCh37"
+                    } else {
+                        "GRCh38"
+                    }
+                ))?),
                 ..Default::default()
             },
             kind,
@@ -436,15 +482,24 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    fn smoke_query_exomes_range_find_none_smaller(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_range_find_none_smaller(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
     ) -> Result<(), anyhow::Error> {
         crate::common::set_snapshot_suffix!("{}-{}-{}", kind, genome_release, version);
         let (common, args, _temp) = build_args(
             ArgsQuery {
-                range: Some(spdi::Range::from_str("GRCh37:1:1:55505598")?),
+                range: Some(spdi::Range::from_str(&format!(
+                    "{}:1:1:55505598",
+                    if genome_release == "grch37" {
+                        "GRCh37"
+                    } else {
+                        "GRCh38"
+                    }
+                ))?),
                 ..Default::default()
             },
             kind,
@@ -460,15 +515,24 @@ mod test {
 
     #[rstest::rstest]
     #[case("exomes", "grch37", "2.1")]
-    fn smoke_query_exomes_range_find_none_larger(
+    #[case("exomes", "grch38", "4.0")]
+    #[case("genomes", "grch38", "4.0")]
+    fn smoke_query_range_find_none_larger(
         #[case] kind: &str,
         #[case] genome_release: &str,
         #[case] version: &str,
     ) -> Result<(), anyhow::Error> {
         crate::common::set_snapshot_suffix!("{}-{}-{}", kind, genome_release, version);
         let (common, args, _temp) = build_args(
             ArgsQuery {
-                range: Some(spdi::Range::from_str("GRCh37:1:55516889:249250621")?),
+                range: Some(spdi::Range::from_str(&format!(
+                    "{}:1:55516889:249250621",
+                    if genome_release == "grch37" {
+                        "GRCh37"
+                    } else {
+                        "GRCh38"
+                    }
+                ))?),
                 ..Default::default()
             },
             kind,

diff --git a/[email protected] → [email protected] b/[email protected] → [email protected]