fix: add variant region condition for dups ending in 3' UTR (#192)

Cargo.toml

@@ -27,7 +27,7 @@ postgres = { version = "0.19", features = ["with-chrono-0_4"] }
 quick_cache = "0.6"
 regex = "1.7"
 rustc-hash = "2.0"
-seqrepo = { version = "0.10.2", features = ["cached"] }
+seqrepo = { version = "0.10.3", features = ["cached"] }
 serde_json = "1.0"
 serde = { version = "1.0", features = ["derive"] }
 thiserror = "1.0"

src/mapper/altseq.rs

@@ -321,7 +321,13 @@ impl AltSeqBuilder {
         match &self.var_c {
             HgvsVariant::CdsVariant { loc_edit, .. } => {
                 let loc = loc_edit.loc.inner();
-                if loc.start.cds_from == CdsFrom::End && loc.end.cds_from == CdsFrom::End {
+                let edit = loc_edit.edit.inner();
+                #[allow(clippy::nonminimal_bool)]
+                if loc.start.cds_from == CdsFrom::End && loc.end.cds_from == CdsFrom::End
+                    || ((edit.is_ins() || edit.is_dup()) && loc.end.cds_from == CdsFrom::End)
+                {
+                    // 3' UTR if both ends are 3' UTR or if the variant is an insertion/duplication
+                    // and the end is 3' UTR.
                     VariantLocation::ThreePrimeUtr
                 } else if loc.start.base < 0 && loc.end.base < 0 {
                     VariantLocation::FivePrimeUtr

src/mapper/variant.rs

@@ -1136,6 +1136,28 @@
         Ok(())
     }
 
+    #[test]
+    fn map_of_ins_three_prime_utr() -> Result<(), Error> {
+        let mapper = build_mapper()?;
+        let hgvs_c = "NM_004985.4:c.567_*1insCCC"; // gene KRAS
+        let var_c = HgvsVariant::from_str(hgvs_c)?;
+        let var_p = mapper.c_to_p(&var_c, None)?;
+        assert_eq!(format!("{}", &var_p), "NP_004976.2:p.?");
+
+        Ok(())
+    }
+
+    #[test]
+    fn map_of_dup_three_prime_utr() -> Result<(), Error> {
+        let mapper = build_mapper()?;
+        let hgvs_c = "NM_153223.3:c.2959_*1dup"; // gene CEP120
+        let var_c = HgvsVariant::from_str(hgvs_c)?;
+        let var_p = mapper.c_to_p(&var_c, None)?;
+        assert_eq!(format!("{}", &var_p), "NP_694955.2:p.?");
+
+        Ok(())
+    }
+
     // TODO(#17): Need to implement validation.
     // #[test]
     // fn map_of_c_out_of_reference_bound() -> Result<(), Error> {
@@ -1622,14 +1644,32 @@
     }
 
     #[test]
-    fn hgvs_c_to_p_three_prime_utr() -> Result<(), Error> {
+    fn hgvs_c_to_p_sub_three_prime_ut() -> Result<(), Error> {
         let hgvsc = "NM_999999.1:c.*3G>A";
         let hgvsp_expected = "MOCK:p.?";
         test_hgvs_c_to_p_conversion(hgvsc, hgvsp_expected)?;
 
         Ok(())
     }
 
+    #[test]
+    fn hgvs_c_to_p_ins_three_prime_utr() -> Result<(), Error> {
+        let hgvsc = "NM_999999.1:c.39_*1insA";
+        let hgvsp_expected = "MOCK:p.?";
+        test_hgvs_c_to_p_conversion(hgvsc, hgvsp_expected)?;
+
+        Ok(())
+    }
+
+    #[test]
+    fn hgvs_c_to_p_dup_three_prime_utr() -> Result<(), Error> {
+        let hgvsc = "NM_999999.1:c.12_*1dup";
+        let hgvsp_expected = "MOCK:p.?";
+        test_hgvs_c_to_p_conversion(hgvsc, hgvsp_expected)?;
+
+        Ok(())
+    }
+
     #[test]
     fn hgvs_c_to_p_deletion_into_three_prime_utr_frameshift() -> Result<(), Error> {
         let hgvsc = "NM_999999.1:c.27_*3del";
@@ -1759,7 +1799,7 @@
             pub hgvs_c: String,
             #[serde(alias = "HGVSp")]
             pub hgvs_p: Option<String>,
             pub description: Option<String>,
             pub alternatives: Option<String>,
         }
 

src/parser/ds.rs

@@ -116,6 +116,16 @@ impl NaEdit {
         }
     }
 
+    /// Returns whether the edit is an insertion.
+    pub fn is_ins(&self) -> bool {
+        matches!(self, NaEdit::Ins { .. })
+    }
+
+    /// Returns whether the edit is a duplication.
+    pub fn is_dup(&self) -> bool {
+        matches!(self, NaEdit::Dup { .. })
+    }
+
     /// Ensures that the reference is a count and no reference bases.
     pub fn with_num(&self) -> Self {
         match self {

tests/data/data/bootstrap.sh

@@ -69,9 +69,9 @@ ADRB2
 AGBL5
 AKR1C3
 ALG9
+ALMS1
 AOAH
 ASB18
-ALMS1
 ATM
 BAHCC1
 BOLA3
@@ -80,6 +80,7 @@ BRCA2
 C18orf32
 CACNA1S
 CCNQ
+CEP120
 COL1A1
 COL3A1
 DEFB133
@@ -95,15 +96,16 @@ FBN1
 FBXO22
 FOLR3
 GTF3C2
-HIST3H2A
 H2AW
 HELQ
 HEXB
+HIST3H2A
 HMGA1
 IFNA16
 IRAK3
 JRK
 KCNIP4
+KRAS
 L1CAM
 LCE2B
 LCE3C
@@ -115,8 +117,8 @@ MED21
 MLH1
 MSH6
 MYH7
-NR1H2
 NEFL
+NR1H2
 OMA1
 OPA1
 OR9A4