Fix: Ensembl ID converstion must be u64 (#618)

Author: Oliver Stolpe

protos/varfish/v1/strucvars/bgdb.proto

@@ -67,7 +67,7 @@ message XlinkRecord {
   // Entrez gene ID.
   uint32 entrez_id = 3;
   // Ensembl gene ID.
-  uint32 ensembl_id = 4;
+  uint64 ensembl_id = 4;
 }
 
 // Gene cross-link database.

src/common/mod.rs

@@ -99,8 +99,8 @@ pub fn reciprocal_overlap(lhs: Range<i32>, rhs: Range<i32>) -> f32 {
     }
 }
 
-/// Helper to convert ENSEMBL and RefSeq gene ID to u32.
-pub fn numeric_gene_id(raw_id: &str) -> Result<u32, anyhow::Error> {
+/// Helper to convert ENSEMBL and RefSeq gene ID to u64.
+pub fn numeric_gene_id(raw_id: &str) -> Result<u64, anyhow::Error> {
     let clean_id = if raw_id.starts_with("ENSG") {
         // Strip "ENSG" prefix and as many zeroes as follow
         raw_id
@@ -113,7 +113,7 @@ pub fn numeric_gene_id(raw_id: &str) -> Result<u32, anyhow::Error> {
     };
 
     clean_id
-        .parse::<u32>()
+        .parse::<u64>()
         .map_err(|e| anyhow::anyhow!("could not parse gene id {:?}: {}", &clean_id, &e))
 }
 
@@ -694,7 +694,7 @@ mod test {
     #[rstest::rstest]
     #[case("ENSG0000000142", 142)]
     #[case("42", 42)]
-    fn numeric_gene_id(#[case] raw_id: &str, #[case] expected: u32) {
+    fn numeric_gene_id(#[case] raw_id: &str, #[case] expected: u64) {
         let actual = super::numeric_gene_id(raw_id).unwrap();
         assert_eq!(expected, actual);
     }

src/strucvars/query/genes.rs

@@ -13,7 +13,7 @@ use crate::{common::GenomeRelease, pbs};
 #[derive(Default, Debug)]
 pub struct XlinkDbRecord {
     pub entrez_id: u32,
-    pub ensembl_gene_id: u32,
+    pub ensembl_gene_id: u64,
     pub symbol: String,
     pub hgnc_id: String,
 }
@@ -26,7 +26,7 @@ pub struct XlinkDb {
     /// Link from entrez ID to indices in records.
     pub from_entrez: multimap::MultiMap<u32, u32>,
     /// Link from ensembl ID to indices in records.
-    pub from_ensembl: multimap::MultiMap<u32, u32>,
+    pub from_ensembl: multimap::MultiMap<u64, u32>,
     /// Link from HGNC ID to indices in records.
     pub from_hgnc: multimap::MultiMap<String, u32>,
 }

src/strucvars/query/mod.rs

@@ -856,7 +856,12 @@ pub fn translate_genes(genes: &Vec<String>, dbs: &InMemoryDbs) -> HashSet<String
         let gene = gene.trim();
         if re_entrez.is_match(gene) {
             if let Ok(gene_id) = numeric_gene_id(gene) {
-                if let Some(record_ids) = dbs.genes.xlink.from_ensembl.get_vec(&gene_id) {
+                if let Some(record_ids) = dbs
+                    .genes
+                    .xlink
+                    .from_entrez
+                    .get_vec(&u32::try_from(gene_id).expect("entrez id should fit in u32"))
+                {
                     for record_id in record_ids {
                         result.insert(dbs.genes.xlink.records[*record_id as usize].hgnc_id.clone());
                     }
@@ -867,7 +872,7 @@ pub fn translate_genes(genes: &Vec<String>, dbs: &InMemoryDbs) -> HashSet<String
             }
         } else if re_ensembl.is_match(gene) {
             if let Ok(gene_id) = numeric_gene_id(gene) {
-                if let Some(record_ids) = dbs.genes.xlink.from_entrez.get_vec(&gene_id) {
+                if let Some(record_ids) = dbs.genes.xlink.from_ensembl.get_vec(&gene_id) {
                     for record_id in record_ids {
                         result.insert(dbs.genes.xlink.records[*record_id as usize].hgnc_id.clone());
                     }