From b0a804d1eb09feb19c6039c7610192f7ef5a5613 Mon Sep 17 00:00:00 2001
From: SimonVandeVyver <simon.vandevyver@ugent.be>
Date: Wed, 11 Sep 2024 09:54:33 +0200
Subject: [PATCH] fix tests to only use characters of peptide alphabet

---
 sa-index/src/sa_searcher.rs | 10 +++++-----
 text-compression/src/lib.rs |  4 ++--
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/sa-index/src/sa_searcher.rs b/sa-index/src/sa_searcher.rs
index 7f60cbb..2324046 100644
--- a/sa-index/src/sa_searcher.rs
+++ b/sa-index/src/sa_searcher.rs
@@ -481,7 +481,7 @@ mod tests {
     }
 
     fn get_example_proteins() -> Proteins {
-        let input_string = "AI-BLACVAA-AC-KCRLZ$";
+        let input_string = "AI-CLACVAA-AC-KCRLY$";
         let text = ProteinText::from_string(input_string);
 
         Proteins {
@@ -561,7 +561,7 @@ mod tests {
         assert_eq!(bounds_res, BoundSearchResult::SearchResult((13, 16)));
 
         // search bounds 'RIZ' with equal I and L
-        let bounds_res = searcher.search_bounds(&[b'R', b'I', b'Z']);
+        let bounds_res = searcher.search_bounds(&[b'R', b'I', b'Y']);
         assert_eq!(bounds_res, BoundSearchResult::SearchResult((17, 18)));
     }
 
@@ -574,18 +574,18 @@ mod tests {
         let searcher = Searcher::new(sa, proteins, Box::new(suffix_index_to_protein));
 
         // search bounds 'RIZ' with equal I and L
-        let found_suffixes = searcher.search_matching_suffixes(&[b'R', b'I', b'Z'], usize::MAX, true);
+        let found_suffixes = searcher.search_matching_suffixes(&[b'R', b'I', b'Y'], usize::MAX, true);
         assert_eq!(found_suffixes, SearchAllSuffixesResult::SearchResult(vec![16]));
 
         // search bounds 'RIZ' without equal I and L
-        let found_suffixes = searcher.search_matching_suffixes(&[b'R', b'I', b'Z'], usize::MAX, false);
+        let found_suffixes = searcher.search_matching_suffixes(&[b'R', b'I', b'Y'], usize::MAX, false);
         assert_eq!(found_suffixes, SearchAllSuffixesResult::NoMatches);
     }
 
     // test edge case where an I or L is the first index in the sparse SA.
     #[test]
     fn test_l_first_index_in_sa() {
-        let input_string = "LMOXZ$";
+        let input_string = "LMPYY$";
         let text = ProteinText::from_string(input_string);
 
         let proteins = Proteins {
diff --git a/text-compression/src/lib.rs b/text-compression/src/lib.rs
index 60b2463..b090826 100644
--- a/text-compression/src/lib.rs
+++ b/text-compression/src/lib.rs
@@ -16,7 +16,7 @@ impl ProteinText {
 
     fn create_char_to_5bit_hashmap() -> HashMap<u8, u8> {
         let mut hashmap = HashMap::<u8, u8>::new();
-        for (i, c) in "ACDEFGHIKLMNPQRSTVWY-".chars().enumerate() {
+        for (i, c) in "ACDEFGHIKLMNPQRSTVWY-$".chars().enumerate() {
             hashmap.insert(c as u8, i as u8);
         }
 
@@ -25,7 +25,7 @@ impl ProteinText {
 
     fn create_bit5_to_char() -> Vec<u8> {
         let mut vec = Vec::<u8>::new();
-        for c in "ACDEFGHIKLMNPQRSTVWY-".chars() {
+        for c in "ACDEFGHIKLMNPQRSTVWY-$".chars() {
             vec.push(c as u8);
         }
         vec