From 7c2971b903bf0bad7a7c28bd5cf29338a8e2dbf1 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Mon, 20 Jan 2025 09:05:06 +0900
Subject: [PATCH 1/3] fix: return no-effect when ter site is inserted around
ter site
---
src/varity/vcf_to_hgvs/protein.clj | 85 ++++++++++++++++--------------
test/varity/vcf_to_hgvs_test.clj | 1 +
2 files changed, 46 insertions(+), 40 deletions(-)
diff --git a/src/varity/vcf_to_hgvs/protein.clj b/src/varity/vcf_to_hgvs/protein.clj
index c4e0d54..bd53fee 100644
--- a/src/varity/vcf_to_hgvs/protein.clj
+++ b/src/varity/vcf_to_hgvs/protein.clj
@@ -179,8 +179,11 @@
(defn- ter-site-same-pos?
[ref-prot-seq alt-prot-seq]
- (let [ter-site-pos (dec (count ref-prot-seq))]
- (= \* (get alt-prot-seq ter-site-pos))))
+ (let [ref-ter-pos (count ref-prot-seq)
+ alt-ter-pos (inc (count (first (string/split alt-prot-seq #"\*"))))]
+ (and (string/includes? ref-prot-seq "*")
+ (string/includes? alt-prot-seq "*")
+ (= ref-ter-pos alt-ter-pos))))
(defn- cds-start-upstream?
[cds-start pos ref alt]
@@ -443,7 +446,9 @@
:else
:frame-shift)
- (and (pos? npref) (= (first palt-only) \*)) :substitution
+ (and (pos? npref) (= (first palt-only) \*)) (if (ter-site-same-pos? ref-prot-seq alt-prot-seq*)
+ :no-effect
+ :substitution)
(not= ref-prot-rest alt-prot-rest) (cond
(or (and (= (first alt-prot-rest) \*)
(>= nprefo npalto)
@@ -572,45 +577,45 @@
(coord/unknown-coordinate))))))))
(defn- protein-extension
- [ppos pref palt {:keys [ref-prot-seq alt-tx-prot-seq c-ter-adjusted-alt-prot-seq ini-offset prefer-extension-for-initial-codon-alt?] :as seq-info}]
- (cond
- (and (not= ppos 1)
- (ter-site-same-pos? ref-prot-seq c-ter-adjusted-alt-prot-seq))
- (mut/protein-no-effect)
+ [ppos pref palt {:keys [ref-prot-seq alt-tx-prot-seq ini-offset prefer-extension-for-initial-codon-alt?] :as seq-info}]
+ (let [alt-prot-seq* (format-alt-prot-seq seq-info)]
+ (cond
+ (and (not= ppos 1)
+ (ter-site-same-pos? ref-prot-seq alt-prot-seq*))
+ (mut/protein-no-effect)
- (and (= ppos 1) (not prefer-extension-for-initial-codon-alt?))
- (mut/protein-unknown-mutation)
+ (and (= ppos 1) (not prefer-extension-for-initial-codon-alt?))
+ (mut/protein-unknown-mutation)
- :else
- (let [[_ ins offset _] (diff-bases (or pref "") (or palt ""))
- alt-prot-seq* (format-alt-prot-seq seq-info)
- ini-site ((comp str first) ref-prot-seq)
- first-diff-aa-info (if (= ppos 1)
- {:ppos 1
- :pref ini-site}
- (get-first-diff-aa-info ppos
- ref-prot-seq
- alt-prot-seq*))
- rest-seq (if (= ppos 1)
- (-> alt-tx-prot-seq
- (subs 0 ini-offset)
- reverse
- (#(apply str %)))
- (subs alt-prot-seq* (:ppos first-diff-aa-info)))
- alt-aa (mut/->long-amino-acid (if (= ppos 1)
- (or (last ins) (first rest-seq))
- (:palt first-diff-aa-info)))
- alt-aa-offset (if (and (= ppos 1) (nil? (last ins))) -1 0)
- new-aa-pos (some-> (string/index-of rest-seq (:pref first-diff-aa-info)) inc (+ alt-aa-offset))]
- (if (and (= ppos 1) (= alt-aa "Ter"))
- (mut/protein-unknown-mutation)
- (mut/protein-extension (if (= ppos 1) (mut/->long-amino-acid ini-site) "Ter")
- (coord/protein-coordinate (if (= ppos 1) 1 (+ ppos offset)))
- alt-aa
- (if (= ppos 1) :upstream :downstream)
- (if new-aa-pos
- (coord/protein-coordinate new-aa-pos)
- (coord/unknown-coordinate)))))))
+ :else
+ (let [[_ ins offset _] (diff-bases (or pref "") (or palt ""))
+ ini-site ((comp str first) ref-prot-seq)
+ first-diff-aa-info (if (= ppos 1)
+ {:ppos 1
+ :pref ini-site}
+ (get-first-diff-aa-info ppos
+ ref-prot-seq
+ alt-prot-seq*))
+ rest-seq (if (= ppos 1)
+ (-> alt-tx-prot-seq
+ (subs 0 ini-offset)
+ reverse
+ (#(apply str %)))
+ (subs alt-prot-seq* (:ppos first-diff-aa-info)))
+ alt-aa (mut/->long-amino-acid (if (= ppos 1)
+ (or (last ins) (first rest-seq))
+ (:palt first-diff-aa-info)))
+ alt-aa-offset (if (and (= ppos 1) (nil? (last ins))) -1 0)
+ new-aa-pos (some-> (string/index-of rest-seq (:pref first-diff-aa-info)) inc (+ alt-aa-offset))]
+ (if (and (= ppos 1) (= alt-aa "Ter"))
+ (mut/protein-unknown-mutation)
+ (mut/protein-extension (if (= ppos 1) (mut/->long-amino-acid ini-site) "Ter")
+ (coord/protein-coordinate (if (= ppos 1) 1 (+ ppos offset)))
+ alt-aa
+ (if (= ppos 1) :upstream :downstream)
+ (if new-aa-pos
+ (coord/protein-coordinate new-aa-pos)
+ (coord/unknown-coordinate))))))))
(defn- protein-indel
[ppos pref palt {:keys [ref-prot-seq c-ter-adjusted-alt-prot-seq
diff --git a/test/varity/vcf_to_hgvs_test.clj b/test/varity/vcf_to_hgvs_test.clj
index b59f4d9..2d667ec 100644
--- a/test/varity/vcf_to_hgvs_test.clj
+++ b/test/varity/vcf_to_hgvs_test.clj
@@ -301,6 +301,7 @@
"chr11" 14279340 "G" "A" '("p.=") ; not actual example (-)
"chr7" 55019277 "G" "GTC" '("p.=") ; not actual example (+)
"chr17" 21042835 "T" "TG" '("p.=") ; not actual example (-)
+ "chr13" 24421121 "A" "ATTA" '("p.=") ; not actual example (-)
;; unknown
"chr12" 40393453 "G" "A" '("p.?") ; not actual example (+)
From 4610278093d9379c0e2c26d8a8acb082275cebc4 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Tue, 21 Jan 2025 15:55:11 +0900
Subject: [PATCH 2/3] fix: fix protein-extension condition
---
src/varity/vcf_to_hgvs/protein.clj | 10 ++++++----
test/varity/vcf_to_hgvs_test.clj | 2 ++
2 files changed, 8 insertions(+), 4 deletions(-)
diff --git a/src/varity/vcf_to_hgvs/protein.clj b/src/varity/vcf_to_hgvs/protein.clj
index bd53fee..f5d5794 100644
--- a/src/varity/vcf_to_hgvs/protein.clj
+++ b/src/varity/vcf_to_hgvs/protein.clj
@@ -428,8 +428,12 @@
pref-only
palt-only)
ini-site-affected (ini-site-affected? ref-exon-seq alt-exon-seq)
+ first-diff-aa-is-ter-site (first-diff-aa-is-ter-site? base-ppos
+ ref-prot-seq
+ alt-prot-seq*)
t (cond
- ref-include-from-ter-start-and-over-ter-end :frame-shift
+ (and ref-include-from-ter-start-and-over-ter-end
+ (not first-diff-aa-is-ter-site)) :frame-shift
(= (+ base-ppos offset) (count ref-prot-seq)) (if (and (= "" pref-only palt-only)
(ter-site-same-pos? ref-prot-seq alt-prot-seq*))
:no-effect
@@ -455,9 +459,7 @@
(= palt (subs pref 0 (count palt))))
(= (first palt-only) \*)) :fs-ter-substitution
ref-include-ter-site :indel
- (first-diff-aa-is-ter-site? base-ppos
- ref-prot-seq
- alt-prot-seq*) :extension
+ first-diff-aa-is-ter-site :extension
:else :frame-shift)
(or (and (zero? nprefo) (zero? npalto))
(and (= nprefo 1) (= npalto 1))) :substitution
diff --git a/test/varity/vcf_to_hgvs_test.clj b/test/varity/vcf_to_hgvs_test.clj
index 2d667ec..2e4c204 100644
--- a/test/varity/vcf_to_hgvs_test.clj
+++ b/test/varity/vcf_to_hgvs_test.clj
@@ -280,6 +280,8 @@
"chr11" 125655318 "TGA" "TAT" '("p.*477Yext*17" "p.*443Yext*17" "p.*477Yext*24")
"chr10" 8074014 "C" "CATGGGTT" '("p.*445Yext*64" "p.*444Yext*64") ; not actual example (+)
"chr10" 87965468 "TC" "T" '("p.*404Eext*11" "p.*577Eext*11" "p.*207Eext*11") ; not actual example (+)
+ "chrX" 15823239 "ATAA" "A" '("p.*483Text*?") ; not actual example (+)
+ "chr13" 24421118 "CTTA" "C" '("p.*1725Vext*2") ; not actual example (-)
;; NOTE: There are very few correct examples...
;; Extension without termination site
From aaf40df800f5f9b8e641a366d6a674e6d2fa01e3 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Tue, 21 Jan 2025 15:56:54 +0900
Subject: [PATCH 3/3] refactor: tweak ter site pos check fn
---
src/varity/vcf_to_hgvs/protein.clj | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/varity/vcf_to_hgvs/protein.clj b/src/varity/vcf_to_hgvs/protein.clj
index f5d5794..2e5b5a4 100644
--- a/src/varity/vcf_to_hgvs/protein.clj
+++ b/src/varity/vcf_to_hgvs/protein.clj
@@ -179,10 +179,10 @@
(defn- ter-site-same-pos?
[ref-prot-seq alt-prot-seq]
- (let [ref-ter-pos (count ref-prot-seq)
- alt-ter-pos (inc (count (first (string/split alt-prot-seq #"\*"))))]
- (and (string/includes? ref-prot-seq "*")
- (string/includes? alt-prot-seq "*")
+ (and (string/includes? ref-prot-seq "*")
+ (string/includes? alt-prot-seq "*")
+ (let [ref-ter-pos (count ref-prot-seq)
+ alt-ter-pos (inc (count (first (string/split alt-prot-seq #"\*"))))]
(= ref-ter-pos alt-ter-pos))))
(defn- cds-start-upstream?