From 654c669e809bb1fe3681e603ba696b6cf15a0857 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Fri, 29 Nov 2024 16:51:55 +0900
Subject: [PATCH] fix: fix protein-insertion regex

---
 src/clj_hgvs/mutation.cljc       |  3 ++-
 test/clj_hgvs/mutation_test.cljc | 13 +++++++++++--
 2 files changed, 13 insertions(+), 3 deletions(-)

diff --git a/src/clj_hgvs/mutation.cljc b/src/clj_hgvs/mutation.cljc
index 89e1b54..752e701 100644
--- a/src/clj_hgvs/mutation.cljc
+++ b/src/clj_hgvs/mutation.cljc
@@ -1760,6 +1760,7 @@
 ;;;
 ;;; e.g. Lys23_Leu24insArgSerGln
 ;;;      Arg78_Gly79insX[23]
+;;;      Thr89_Ter90insSerProIle (T89_*90insSPI)
 
 (defrecord ProteinInsertion [ref-start coord-start ref-end coord-end alts]
   Mutation
@@ -1819,7 +1820,7 @@
                     vec)))
 
 (def ^:private protein-insertion-re
-  #"([A-Z](?:[a-z]{2})?)(\d+)_([A-Z](?:[a-z]{2})?)(\d+)ins([\da-zA-Z*\[\]]+)")
+  #"([A-Z](?:[a-z]{2})?)(\d+)_([A-Z*](?:[a-z]{2})?)(\d+)ins([\da-zA-Z*\[\]]+)")
 
 (defn parse-protein-insertion
   [s]
diff --git a/test/clj_hgvs/mutation_test.cljc b/test/clj_hgvs/mutation_test.cljc
index 85061d7..593aadc 100644
--- a/test/clj_hgvs/mutation_test.cljc
+++ b/test/clj_hgvs/mutation_test.cljc
@@ -1371,6 +1371,11 @@
 (def protein-insertion4 (mut/protein-insertion "Arg" (coord/protein-coordinate 78)
                                                "Gly" (coord/protein-coordinate 79)
                                                ["Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa"]))
+(def protein-insertion5s "Thr89_Ter90insSerProIle")
+(def protein-insertion5ss "T89_*90insSPI")
+(def protein-insertion5 (mut/protein-insertion "Thr" (coord/protein-coordinate 89)
+                                               "Ter" (coord/protein-coordinate 90)
+                                               ["Ser" "Pro" "Ile"]))
 
 (deftest format-protein-insertion-test
   (testing "returns a string expression of a protein insertion"
@@ -1386,7 +1391,9 @@
       protein-insertion3 {:ins-format :count} protein-insertion3sc
       protein-insertion3 {:amino-acid-format :short :ins-format :count} protein-insertion3scs
       protein-insertion4 {:ins-format :auto} protein-insertion4sc
-      protein-insertion4 {:amino-acid-format :short :ins-format :auto} protein-insertion4scs)))
+      protein-insertion4 {:amino-acid-format :short :ins-format :auto} protein-insertion4scs
+      protein-insertion5 nil protein-insertion5s
+      protein-insertion5 {:amino-acid-format :short} protein-insertion5ss)))
 
 (deftest parse-protein-insertion-test
   (testing "returns a correct ProteinInsertion"
@@ -1400,7 +1407,9 @@
       protein-insertion3sc protein-insertion3
       protein-insertion3scs protein-insertion3
       protein-insertion4sc protein-insertion4
-      protein-insertion4scs protein-insertion4))
+      protein-insertion4scs protein-insertion4
+      protein-insertion5s protein-insertion5
+      protein-insertion5ss protein-insertion5))
   (testing "invalid protein insertion"
     (are [s] (thrown? #?(:clj Throwable, :cljs js/Error)
                       (mut/parse-protein-insertion s))