From cad3fb5967d28e8d2035a6dbd6aaf704bfaf7ff0 Mon Sep 17 00:00:00 2001
From: tibvdm <tibovandemoortele@hotmail.com>
Date: Wed, 28 Aug 2024 20:31:37 +0200
Subject: [PATCH] return references of the proteins instead of cloning the
 entire struct

---
 sa-index/src/peptide_search.rs | 14 +++++++-------
 sa-mappings/Cargo.toml         |  2 ++
 sa-mappings/src/proteins.rs    |  2 ++
 3 files changed, 11 insertions(+), 7 deletions(-)

diff --git a/sa-index/src/peptide_search.rs b/sa-index/src/peptide_search.rs
index 5f4d8b3..47ff3e1 100644
--- a/sa-index/src/peptide_search.rs
+++ b/sa-index/src/peptide_search.rs
@@ -5,9 +5,9 @@ use serde::Serialize;
 use crate::sa_searcher::{SearchAllSuffixesResult, Searcher};
 
 #[derive(Debug, Serialize)]
-pub struct SearchResult {
+pub struct SearchResult<'a> {
     pub sequence: String,
-    pub proteins: Vec<ProteinInfo>,
+    pub proteins: Vec<&'a Protein>,
     pub cutoff_used: bool
 }
 
@@ -75,12 +75,12 @@ pub fn search_proteins_for_peptide<'a>(
     Some((cutoff_used, proteins))
 }
 
-pub fn search_peptide(searcher: &Searcher, peptide: &str, cutoff: usize, equate_il: bool) -> Option<SearchResult> {
+pub fn search_peptide<'a>(searcher: &'a Searcher, peptide: &str, cutoff: usize, equate_il: bool) -> Option<SearchResult<'a>> {
     let (cutoff_used, proteins) = search_proteins_for_peptide(searcher, peptide, cutoff, equate_il)?;
 
     Some(SearchResult {
         sequence: peptide.to_string(),
-        proteins: proteins.iter().map(|&protein| protein.into()).collect(),
+        proteins,
         cutoff_used
     })
 }
@@ -99,12 +99,12 @@ pub fn search_peptide(searcher: &Searcher, peptide: &str, cutoff: usize, equate_
 /// # Returns
 ///
 /// Returns an `OutputData<SearchOnlyResult>` object with the search results for the peptides
-pub fn search_all_peptides(
-    searcher: &Searcher,
+pub fn search_all_peptides<'a>(
+    searcher: &'a Searcher,
     peptides: &Vec<String>,
     cutoff: usize,
     equate_il: bool
-) -> Vec<SearchResult> {
+) -> Vec<SearchResult<'a>> {
     peptides
         .par_iter()
         .filter_map(|peptide| search_peptide(searcher, peptide, cutoff, equate_il))
diff --git a/sa-mappings/Cargo.toml b/sa-mappings/Cargo.toml
index b20a2bf..69d13c6 100644
--- a/sa-mappings/Cargo.toml
+++ b/sa-mappings/Cargo.toml
@@ -11,3 +11,5 @@ tempdir = "0.3.7"
 [dependencies]
 fa-compression = { path = "../fa-compression" }
 bytelines = "2.5.0"
+serde = { version = "1.0.197", features = ["derive"] }
+serde_json = "1.0.116"
diff --git a/sa-mappings/src/proteins.rs b/sa-mappings/src/proteins.rs
index 7597b8d..35777c9 100644
--- a/sa-mappings/src/proteins.rs
+++ b/sa-mappings/src/proteins.rs
@@ -5,6 +5,7 @@ use std::{error::Error, fs::File, io::BufReader, ops::Index, str::from_utf8};
 
 use bytelines::ByteLines;
 use fa_compression::algorithm1::{decode, encode};
+use serde::Serialize;
 
 /// The separation character used in the input string
 pub static SEPARATION_CHARACTER: u8 = b'-';
@@ -13,6 +14,7 @@ pub static SEPARATION_CHARACTER: u8 = b'-';
 /// This character should be smaller than the separation character
 pub static TERMINATION_CHARACTER: u8 = b'$';
 
+#[derive(Debug, Clone, Serialize)]
 /// A struct that represents a protein and its linked information
 pub struct Protein {
     /// The id of the protein