feat: adding /genes/lookup endpoint (#193)

src/server/actix_server/genes_lookup.rs

@@ -0,0 +1,69 @@
+//! Implementation of `/genes/lookup` that allows to lookup genes by symbol or identifier.
+//!
+//! In contrast to gene search, more than one query may be given but this must match exactly
+//! the symbol or HGNC/NCBI/ENSEMBL identifier.
+use actix_web::{
+    get,
+    web::{self, Data, Json, Path},
+    Responder,
+};
+
+use crate::server::GeneNames;
+
+use super::error::CustomError;
+use serde_with::{formats::CommaSeparator, StringWithSeparator};
+
+/// Parameters for `handle`.
+#[serde_with::skip_serializing_none]
+#[serde_with::serde_as]
+#[derive(serde::Serialize, serde::Deserialize, Debug, Clone)]
+#[serde(rename_all = "snake_case")]
+struct Request {
+    /// The strings to search for.
+    #[serde_as(as = "StringWithSeparator::<CommaSeparator, String>")]
+    pub q: Vec<String>,
+}
+
+/// Result for `handle`.
+#[derive(serde::Serialize, serde::Deserialize, Debug, Clone)]
+#[serde_with::skip_serializing_none]
+struct Container {
+    // TODO: add data version
+    /// The resulting gene information.
+    pub genes: indexmap::IndexMap<String, Option<GeneNames>>,
+}
+
+/// Query for annotations for one variant.
+#[allow(clippy::option_map_unit_fn)]
+#[get("/genes/lookup")]
+async fn handle(
+    data: Data<crate::server::WebServerData>,
+    _path: Path<()>,
+    query: web::Query<Request>,
+) -> actix_web::Result<impl Responder, CustomError> {
+    if query.q.len() < 2 {
+        return Ok(Json(Container {
+            // server_version: VERSION.to_string(),
+            // builder_version,
+            genes: Default::default(),
+        }));
+    }
+
+    let genes_db = data.genes.as_ref().ok_or(CustomError::new(anyhow::anyhow!(
+        "genes database not available"
+    )))?;
+
+    let genes = indexmap::IndexMap::from_iter(query.q.iter().map(|q| {
+        let v = genes_db
+            .name_to_hgnc_idx
+            .get(q)
+            .map(|idx| genes_db.gene_names[*idx].clone());
+        (q.clone(), v)
+    }));
+
+    Ok(Json(Container {
+        // server_version: VERSION.to_string(),
+        // builder_version,
+        genes,
+    }))
+}

src/server/actix_server/genes_search.rs

@@ -1,4 +1,7 @@
 //! Implementation of `/genes/search` that allows to search for genes by symbol etc.
+//!
+//! Gene identifiers (HGNC, NCBI, ENSEMBL) must match.  As for symbols and names, the
+//! search string may also be a substring.
 use actix_web::{
     get,
     web::{self, Data, Json, Path},
@@ -105,7 +108,7 @@ async fn handle(
     let fields_contains = |field: &Fields| -> bool { fields.is_empty() || fields.contains(field) };
 
     let mut genes = genes_db
-        .gene_strings
+        .gene_names
         .iter()
         .map(|gn| -> Scored<GeneNames> {
             let score = if (fields_contains(&Fields::HgncId) && &gn.hgnc_id == q)

src/server/actix_server/mod.rs

@@ -6,6 +6,7 @@ pub mod annos_variant;
 pub mod error;
 pub mod fetch;
 pub mod genes_info;
+pub mod genes_lookup;
 pub mod genes_search;
 
 use actix_web::{middleware::Logger, web::Data, App, HttpServer};
@@ -26,7 +27,8 @@ pub async fn main(args: &Args, dbs: Data<WebServerData>) -> std::io::Result<()>
             .service(annos_range::handle)
             .service(annos_db_info::handle)
             .service(genes_info::handle)
-            .service(genes_search::handle);
+            .service(genes_search::handle)
+            .service(genes_lookup::handle);
         app.wrap(Logger::default())
     })
     .bind((args.listen_host.as_str(), args.listen_port))?

src/server/mod.rs

@@ -2,7 +2,7 @@
 
 pub mod actix_server;
 
-use std::time::Instant;
+use std::{collections::HashMap, time::Instant};
 
 use clap::Parser;
 use indicatif::ParallelProgressIterator;
@@ -121,7 +121,9 @@ pub struct GeneInfoDb {
     /// The database.
     pub db: rocksdb::DBWithThreadMode<rocksdb::MultiThreaded>,
     /// Gene information to keep in memory (for `/genes/search`).
-    pub gene_strings: Vec<GeneNames>,
+    pub gene_names: Vec<GeneNames>,
+    /// Mapping from allowed gene name string to index in `gene_names`.
+    pub name_to_hgnc_idx: HashMap<String, usize>,
 }
 
 /// Genome-release specific annotation for each database.
@@ -316,18 +318,33 @@ pub fn run(args_common: &common::cli::Args, args: &Args) -> Result<(), anyhow::E
     if let Some(path_genes) = args.path_genes.as_ref() {
         tracing::info!("Opening genes database {}...", path_genes);
         let before_open = Instant::now();
-        let genes = open_db(path_genes, "genes")?;
+        let db = open_db(path_genes, "genes")?;
         tracing::info!(
             "...done opening genes database in {:?}",
             before_open.elapsed()
         );
         tracing::info!("Building gene names...");
         let before_open = Instant::now();
-        let gene_names = extract_gene_names(&genes)?;
+        let gene_names = extract_gene_names(&db)?;
+        let name_to_hgnc_idx = {
+            let mut result = HashMap::new();
+            for (idx, gene_name) in gene_names.iter().enumerate() {
+                result.insert(gene_name.hgnc_id.clone(), idx);
+                if let Some(ensembl_gene_id) = gene_name.ensembl_gene_id.as_ref() {
+                    result.insert(ensembl_gene_id.clone(), idx);
+                }
+                if let Some(ncbi_gene_id) = gene_name.ncbi_gene_id.as_ref() {
+                    result.insert(ncbi_gene_id.clone(), idx);
+                }
+                result.insert(gene_name.symbol.clone(), idx);
+            }
+            result
+        };
         tracing::info!("...done building genes names {:?}", before_open.elapsed());
         data.genes = Some(GeneInfoDb {
-            db: genes,
-            gene_strings: gene_names,
+            db,
+            gene_names,
+            name_to_hgnc_idx,
         });
     }
     // Argument lists from the command line with the corresponding database enum value.
@@ -394,6 +411,11 @@ pub fn run(args_common: &common::cli::Args, args: &Args) -> Result<(), anyhow::E
         args.listen_host.as_str(),
         args.listen_port
     );
+    tracing::info!(
+        "  try: http://{}:{}/genes/lookup?q=BRCA,BRCA1,HGNC:1100",
+        args.listen_host.as_str(),
+        args.listen_port
+    );
     tracing::info!(
         "  try: http://{}:{}/genes/info?hgnc_id=HGNC:12403",
         args.listen_host.as_str(),