feat: initial implementation of ACMG rules for DEL/loss (#1) (#9)

.gitattributes

@@ -0,0 +1,4 @@
+tests/**/hgnc.tsv filter=lfs diff=lfs merge=lfs -text
+tests/**/*.tsv filter=lfs diff=lfs merge=lfs -text
+tests/**/*.bin* filter=lfs diff=lfs merge=lfs -text
+tests/**/rocksdb/** filter=lfs diff=lfs merge=lfs -text

.github/workflows/rust.yml

@@ -22,6 +22,11 @@ jobs:
           override: true
           components: rustfmt
 
+      - name: Setup protoc
+        uses: arduino/[email protected]
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+
       - name: Check format
         run: |
           cargo fmt -- --check
@@ -39,6 +44,11 @@ jobs:
           override: true
           components: clippy
 
+      - name: Setup protoc
+        uses: arduino/[email protected]
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+
       - name: Lint with clippy
         uses: actions-rs/clippy-check@v1
         with:
@@ -59,6 +69,11 @@ jobs:
           toolchain: stable
           override: true
 
+      - name: Setup protoc
+        uses: arduino/[email protected]
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+
       - uses: Swatinem/rust-cache@v2
 
       - name: Run cargo-tarpaulin

Cargo.toml

@@ -13,21 +13,39 @@ name = "scarus"
 path = "src/main.rs"
 
 [dependencies]
+annonars = "0.24.5"
 anyhow = "1.0"
+bio = "1.4.0"
+biocommons-bioutils = "0.1.4"
 clap-verbosity-flag = "2.0"
 clap = { version = "4.4", features = ["derive", "help"] }
+csv = "1.3.0"
+displaydoc = "0.2.4"
+hgvs = "0.13.2"
+itertools = "0.11.0"
 log = "0.4"
+mehari = "0.17"
+rustc-hash = "1.1"
+rustix = "=0.38.21"  # TODO: remove pin
+serde_json = "1.0"
+serde = { version = "1.0", features = ["serde_derive"] }
+serde_with = { version = "3.3", features = ["indexmap_2"] }
 thiserror = "1.0"
 tokio = { version = "1.33", features = ["full"] }
 tracing = "0.1"
 tracing-subscriber = "0.3"
+rocksdb = { version = "0.21.0", features = ["multi-threaded-cf"] }
+prost = "0.12.1"
 
 [dev-dependencies]
 anyhow = "1.0"
+assert-eq-float = "0.1.3"
 clap-verbosity-flag = {version = "2.0"}
 clap = { version = "4.1", features = ["derive", "env"] }
 env_logger = "0.10"
+insta = { version = "1.34", features = ["yaml"] }
 pretty_assertions = "1.3"
+rstest = "0.18.2"
+serde_test = "1.0"
 temp_testdir = "0.2"
-test-log = "0.2"
-tracing-subscriber = {version = "0.3" }
+tracing-test = "0.2"

src/common/mod.rs

@@ -10,3 +10,44 @@ pub struct Args {
     #[clap(flatten)]
     pub verbose: Verbosity<InfoLevel>,
 }
+
+/// Assembly to be passed on the command line.
+#[derive(
+    Debug,
+    Clone,
+    Copy,
+    PartialEq,
+    Eq,
+    PartialOrd,
+    Ord,
+    clap::ValueEnum,
+    serde::Deserialize,
+    serde::Serialize,
+)]
+pub enum Assembly {
+    /// GRCh37
+    Grch37,
+    /// GRCh38
+    Grch38,
+}
+
+impl From<Assembly> for biocommons_bioutils::assemblies::Assembly {
+    fn from(val: Assembly) -> Self {
+        match val {
+            // GRCh37p10 is the first one with chrMT.
+            Assembly::Grch37 => biocommons_bioutils::assemblies::Assembly::Grch37p10,
+            // All canonical contigs including chrMT are in GRCh38.
+            Assembly::Grch38 => biocommons_bioutils::assemblies::Assembly::Grch38,
+        }
+    }
+}
+
+impl From<biocommons_bioutils::assemblies::Assembly> for Assembly {
+    fn from(val: biocommons_bioutils::assemblies::Assembly) -> Self {
+        match val {
+            biocommons_bioutils::assemblies::Assembly::Grch37 => Assembly::Grch37,
+            biocommons_bioutils::assemblies::Assembly::Grch37p10 => Assembly::Grch37,
+            biocommons_bioutils::assemblies::Assembly::Grch38 => Assembly::Grch38,
+        }
+    }
+}

src/main.rs

@@ -1,9 +1,9 @@
 //! Main entry point for Scarus application.
 
-#![deny(clippy::pedantic)]
+// #![deny(clippy::pedantic)]
 #![allow(clippy::must_use_candidate)]
 #![allow(clippy::module_name_repetitions)]
-#![warn(missing_docs)]
+// #![warn(missing_docs)]
 
 use clap::{Parser, Subcommand};