From 0d7b6d8f7ca62b7152dad0d2da6557bb62580e58 Mon Sep 17 00:00:00 2001
From: Manuel Holtgrewe <manuel.holtgrewe@bih-charite.de>
Date: Wed, 18 Oct 2023 10:11:09 +0200
Subject: [PATCH] feat: bump annonars, write out VCV+RCV+clinsig (#217)

---
 Cargo.lock                                    |  4 +-
 Cargo.toml                                    |  2 +-
 src/annotate/seqvars/mod.rs                   | 50 +++++++++++++------
 .../seqvar_freqs/db-rs1263393206/output.vcf   |  4 +-
 4 files changed, 40 insertions(+), 20 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 115be7da..cf9c2376 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -255,9 +255,9 @@ dependencies = [
 
 [[package]]
 name = "annonars"
-version = "0.23.1"
+version = "0.24.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "17a6e189155aa6d3488e373db9625fc80733ec50cda565e6ed4b160a337760c8"
+checksum = "ac9a747c515e24fed680d9a05112c0535e069ff5705bf51256474b47ef66a382"
 dependencies = [
  "actix-web",
  "anyhow",
diff --git a/Cargo.toml b/Cargo.toml
index 36e1badd..fa9b3752 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -27,7 +27,7 @@ path = "src/main.rs"
 
 [dependencies]
 actix-web = "4.4"
-annonars = "0.23"
+annonars = "0.24"
 anyhow = "1.0"
 bgzip = "0.3"
 bio = "1.3"
diff --git a/src/annotate/seqvars/mod.rs b/src/annotate/seqvars/mod.rs
index e9e5ad82..3ae87053 100644
--- a/src/annotate/seqvars/mod.rs
+++ b/src/annotate/seqvars/mod.rs
@@ -216,12 +216,20 @@ fn build_header(header_in: &VcfHeader) -> VcfHeader {
         Map::<Info>::new(
             Number::Count(1),
             Type::String,
-            "ClinVar clinical significance",
+            "ClinVar clinical significance (one highest significance per VCV)",
         ),
     );
     header_out.infos_mut().insert(
         field::Key::from_str("clinvar_rcv").unwrap(),
-        Map::<Info>::new(Number::Count(1), Type::String, "ClinVar RCV accession"),
+        Map::<Info>::new(
+            Number::Count(1),
+            Type::String,
+            "ClinVar RCV accession (corresponds to clinvar_clinsig)",
+        ),
+    );
+    header_out.infos_mut().insert(
+        field::Key::from_str("clinvar_vcv").unwrap(),
+        Map::<Info>::new(Number::Count(1), Type::String, "ClinVar VCV accession"),
     );
 
     header_out
@@ -396,21 +404,33 @@ where
             clinvar_minimal::pbs::Record::decode(&mut std::io::Cursor::new(&raw_value))?;
 
         let clinvar_minimal::pbs::Record {
-            clinical_significance,
-            rcv,
+            vcv,
+            reference_assertions,
             ..
         } = clinvar_record;
-        let clinical_significance: clinvar_minimal::cli::reading::ClinicalSignificance =
-            clinical_significance.into();
-
-        vcf_record.info_mut().insert(
-            field::Key::from_str("clinvar_clinsig").unwrap(),
-            Some(field::Value::String(clinical_significance.to_string())),
-        );
-        vcf_record.info_mut().insert(
-            field::Key::from_str("clinvar_rcv").unwrap(),
-            Some(field::Value::String(rcv)),
-        );
+        if let Some(reference_assertion) = reference_assertions.first() {
+            let clinvar_minimal::pbs::ReferenceAssertion {
+                rcv,
+                clinical_significance,
+                ..
+            } = reference_assertion.clone();
+
+            let clinical_significance: clinvar_minimal::cli::reading::ClinicalSignificance =
+                clinical_significance.into();
+
+            vcf_record.info_mut().insert(
+                field::Key::from_str("clinvar_clinsig").unwrap(),
+                Some(field::Value::String(clinical_significance.to_string())),
+            );
+            vcf_record.info_mut().insert(
+                field::Key::from_str("clinvar_rcv").unwrap(),
+                Some(field::Value::String(rcv)),
+            );
+            vcf_record.info_mut().insert(
+                field::Key::from_str("clinvar_vcv").unwrap(),
+                Some(field::Value::String(vcv)),
+            );
+        }
     }
 
     Ok(())
diff --git a/tests/data/db/create/seqvar_freqs/db-rs1263393206/output.vcf b/tests/data/db/create/seqvar_freqs/db-rs1263393206/output.vcf
index 1237087c..bd04bcf2 100644
--- a/tests/data/db/create/seqvar_freqs/db-rs1263393206/output.vcf
+++ b/tests/data/db/create/seqvar_freqs/db-rs1263393206/output.vcf
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6f3cc926773a089a9ef59a593c5d2d88ad0e639e8b91c43079716e4ac7d26e62
-size 17901
+oid sha256:a901a01e76edcd18dfb20209432b715c5558b2dbe41b1d8063588fd9ba58a330
+size 18050