From ed67196074154013af96a40a1610a7981200087e Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Wed, 25 Sep 2024 08:25:48 +0900
Subject: [PATCH 1/6] fix!: fix dna ins/delins unknown repeat notation
---
src/clj_hgvs/mutation.cljc | 44 +++++++++++++++++++++++---------
test/clj_hgvs/mutation_test.cljc | 19 +++++++++++---
2 files changed, 48 insertions(+), 15 deletions(-)
diff --git a/src/clj_hgvs/mutation.cljc b/src/clj_hgvs/mutation.cljc
index 5e2bd76..03d750d 100644
--- a/src/clj_hgvs/mutation.cljc
+++ b/src/clj_hgvs/mutation.cljc
@@ -430,7 +430,7 @@
;;; g.122_123ins123_234inv (TODO)
;;; g.122_123ins213_234invinsAins123_211inv (TODO)
;;; g.549_550insN
-;;; g.1134_1135ins(100)
+;;; g.1134_1135insN[100]
;;; g.?_?insNC_000023.10:(12345_23456)_(34567_45678)
(defrecord DNAInsertion [coord-start coord-end alt]
@@ -445,10 +445,10 @@
(string? alt) (case ins-format
:auto (if (and (every? #(= % \N) alt)
(>= (count alt) 10))
- (str "(" (count alt) ")")
+ (str "N[" (count alt) "]")
alt)
:bases alt
- :count (str "(" (count alt) ")"))
+ :count (str "N[" (count alt) "]"))
(map? alt) [(:transcript alt)
":"
(coord/format (:coord-start alt))
@@ -478,7 +478,7 @@
(defn- parse-dna-insertion-alt
[s kind]
(or (re-matches #"[A-Z]+" s)
- (some-> (re-matches #"\((\d+)\)" s)
+ (some-> (re-matches #"N\[(\d+)\]" s)
(second)
(intl/parse-long)
(repeat "N")
@@ -624,20 +624,28 @@
;;; e.g. g.6775delinsGA
;;; g.6775delTinsGA
;;; c.145_147delinsTGG
+;;; c.145_147delinsN[10]
(defrecord DNAIndel [coord-start coord-end ref alt]
Mutation
(format [this] (format this nil))
- (format [this {:keys [show-bases?] :or {show-bases? false}}]
+ (format [this {:keys [show-bases? ins-format] :or {show-bases? false ins-format :auto}}]
(apply str (flatten [(coord/format coord-start)
- (if (and coord-end
- (or (not (coord/comparable-coordinates? coord-start coord-end))
- (neg? (compare coord-start coord-end))))
+ (when (and coord-end
+ (or (not (coord/comparable-coordinates? coord-start coord-end))
+ (neg? (compare coord-start coord-end))))
["_" (coord/format coord-end)])
"del"
- (if show-bases? ref)
+ (when show-bases? ref)
"ins"
- alt])))
+ (when (string? alt)
+ (case ins-format
+ :auto (if (and (every? #(= % \N) alt)
+ (>= (count alt) 10))
+ (str "N[" (count alt) "]")
+ alt)
+ :bases alt
+ :count (str "N[" (count alt) "]")))])))
(plain [this]
(into {:mutation "dna-indel"} (plain-coords this)))
@@ -663,6 +671,15 @@
:clj-hgvs.mutation.dna-indel/ref
:clj-hgvs.mutation.dna-indel/alt])))
+(defn- parse-dna-indel-alt
+ [s]
+ (or (re-matches #"[A-Z]+" s)
+ (some-> (re-matches #"N\[(\d+)\]" s)
+ (second)
+ (intl/parse-long)
+ (repeat "N")
+ (#(apply str %)))))
+
(defn dna-indel
"Constructor of DNAIndel. Throws an exception if any input is illegal."
[coord-start coord-end ref alt]
@@ -673,13 +690,16 @@
(DNAIndel. coord-start coord-end ref alt))
(def ^:private dna-indel-re
- #"([\d\-\+\*\?]+)(?:_([\d\-\+\*\?]+))?del([A-Z]+)?ins([A-Z]+)")
+ #"([\d\-\+\*\?]+)(?:_([\d\-\+\*\?]+))?del([A-Z]+)?ins(N\[\d+\]|[A-Z]+)")
(defn parse-dna-indel
[s kind]
(let [[_ coord-s coord-e ref alt] (re-matches dna-indel-re s)
parse-coord (coord-parser kind)]
- (dna-indel (parse-coord coord-s) (some-> coord-e parse-coord) ref alt)))
+ (dna-indel (parse-coord coord-s)
+ (some-> coord-e parse-coord)
+ ref
+ (parse-dna-indel-alt alt))))
(defmethod restore "dna-indel"
[m]
diff --git a/test/clj_hgvs/mutation_test.cljc b/test/clj_hgvs/mutation_test.cljc
index a49a292..28eb922 100644
--- a/test/clj_hgvs/mutation_test.cljc
+++ b/test/clj_hgvs/mutation_test.cljc
@@ -342,7 +342,7 @@
"N"))
(def dna-insertion6sb "1134_1135insNNNNNNNNNN")
-(def dna-insertion6sc "1134_1135ins(10)")
+(def dna-insertion6sc "1134_1135insN[10]")
(def dna-insertion6k :genome)
(def dna-insertion6 (mut/dna-insertion (coord/genomic-coordinate 1134)
(coord/genomic-coordinate 1135)
@@ -524,20 +524,33 @@
nil
"TGG"))
+(def dna-indel4sb "1134_1138delinsNNNNNNNNNN")
+(def dna-indel4sc "1134_1138delinsN[10]")
+(def dna-indel4k :genome)
+(def dna-indel4 (mut/dna-indel (coord/genomic-coordinate 1134)
+ (coord/genomic-coordinate 1138)
+ nil
+ "NNNNNNNNNN"))
+
(deftest format-dna-indel-test
(testing "returns a string expression of a DNA indel"
(are [m o s] (= (mut/format m o) s)
dna-indel1 nil dna-indel1s
dna-indel2 nil dna-indel2ss
dna-indel2 {:show-bases? true} dna-indel2s
- dna-indel3 nil dna-indel3s)))
+ dna-indel3 nil dna-indel3s
+ dna-indel4 {:ins-format :auto} dna-indel4sc
+ dna-indel4 {:ins-format :bases} dna-indel4sb
+ dna-indel4 {:ins-format :count} dna-indel4sc)))
(deftest parse-dna-indel-test
(testing "returns a correct DNAIndel"
(are [s k m] (= (mut/parse-dna-indel s k) m)
dna-indel1s dna-indel1k dna-indel1
dna-indel2s dna-indel2k dna-indel2
- dna-indel3s dna-indel3k dna-indel3))
+ dna-indel3s dna-indel3k dna-indel3
+ dna-indel4sb dna-indel4k dna-indel4
+ dna-indel4sc dna-indel4k dna-indel4))
(testing "invalid DNA indel"
(are [s k] (thrown? #?(:clj Throwable, :cljs js/Error)
(mut/parse-dna-indel s k))
From c2e025d6291b0ca75b473093c64cfdc16598d8ad Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Wed, 25 Sep 2024 08:50:18 +0900
Subject: [PATCH 2/6] fix!: fix rna ins unknown repeat notation
---
src/clj_hgvs/mutation.cljc | 12 ++++++------
test/clj_hgvs/mutation_test.cljc | 2 +-
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/src/clj_hgvs/mutation.cljc b/src/clj_hgvs/mutation.cljc
index 03d750d..adf6222 100644
--- a/src/clj_hgvs/mutation.cljc
+++ b/src/clj_hgvs/mutation.cljc
@@ -1056,7 +1056,7 @@
;;; RNA - insertion
;;;
;;; e.g. r.756_757insacu
-;;; r.431_432ins(5)
+;;; r.431_432insn[5]
;;; r.123_124insL37425.1:23_361
(defrecord RNAInsertion [coord-start coord-end alt]
@@ -1069,7 +1069,7 @@
"ins"
(cond
(map? alt) (str (:genbank alt) ":" (:coord-start alt) "_" (:coord-end alt))
- (re-matches #"n{2,}" alt) (str "(" (count alt) ")")
+ (re-matches #"n{2,}" alt) (str "n[" (count alt) "]")
:else alt)))
(plain [this]
(into {:mutation "rna-insertion"} (plain-coords this))))
@@ -1094,9 +1094,9 @@
(defn- parse-rna-alt-n
[s]
- (if-let [n (some-> (re-find #"\((\d)\)" s)
- second
- intl/parse-long)]
+ (when-let [n (some-> (re-find #"n\[(\d+)\]" s)
+ second
+ intl/parse-long)]
(apply str (repeat n "n"))))
(def ^:private genbank-re
@@ -1118,8 +1118,8 @@
(rna-insertion (coord/parse-rna-coordinate coord-s)
(some-> coord-e coord/parse-rna-coordinate)
(cond
+ (re-find #"n\[\d+\]" alt) (parse-rna-alt-n alt)
(re-find #"[a-z]+" alt) alt
- (re-find #"\(\d\)" alt) (parse-rna-alt-n alt)
:else (parse-rna-alt-genbank alt)))))
(defmethod restore "rna-insertion"
diff --git a/test/clj_hgvs/mutation_test.cljc b/test/clj_hgvs/mutation_test.cljc
index 28eb922..d83482f 100644
--- a/test/clj_hgvs/mutation_test.cljc
+++ b/test/clj_hgvs/mutation_test.cljc
@@ -831,7 +831,7 @@
:coord-end (coord/plain (coord/rna-coordinate 757 nil nil))
:alt "acu"})
-(def rna-insertion2s "431_432ins(5)")
+(def rna-insertion2s "431_432insn[5]")
(def rna-insertion2 (mut/rna-insertion (coord/rna-coordinate 431 nil nil)
(coord/rna-coordinate 432 nil nil)
"nnnnn"))
From e11455b174522fe97c282ecac50b57aa99413826 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Wed, 25 Sep 2024 09:04:09 +0900
Subject: [PATCH 3/6] fix!: fix protein ins unknown repeat notation
---
src/clj_hgvs/mutation.cljc | 12 ++++++++----
test/clj_hgvs/mutation_test.cljc | 2 +-
2 files changed, 9 insertions(+), 5 deletions(-)
diff --git a/src/clj_hgvs/mutation.cljc b/src/clj_hgvs/mutation.cljc
index adf6222..7041543 100644
--- a/src/clj_hgvs/mutation.cljc
+++ b/src/clj_hgvs/mutation.cljc
@@ -1744,7 +1744,7 @@
;;; Protein - insertion
;;;
;;; e.g. Lys23_Leu24insArgSerGln
-;;; Arg78_Gly79ins23
+;;; Arg78_Gly79insX[23]
(defrecord ProteinInsertion [ref-start coord-start ref-end coord-end alts]
Mutation
@@ -1759,7 +1759,7 @@
(coord/format coord-end)
"ins"
(if (every? #(= % "Xaa") alts)
- (count alts)
+ (str "X[" (count alts) "]")
(cond->> alts
(= amino-acid-format :short) (map ->short-amino-acid)))])))
(plain [this]
@@ -1790,10 +1790,14 @@
[s]
(condp re-matches s
#"([A-Z*]([a-z]{2})?)+" (mapv ->long-amino-acid (re-seq #"[A-Z*](?:[a-z]{2})?" s))
- #"\d+" (vec (repeat (intl/parse-long s) "Xaa"))))
+ #"X\[\d+\]" (-> (re-find #"X\[(\d+)\]" s)
+ second
+ intl/parse-long
+ (repeat "Xaa")
+ 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 d83482f..aa1e52d 100644
--- a/test/clj_hgvs/mutation_test.cljc
+++ b/test/clj_hgvs/mutation_test.cljc
@@ -1327,7 +1327,7 @@
"Leu" (coord/protein-coordinate 24)
["Arg" "Ser" "Ter"]))
-(def protein-insertion3s "Arg78_Gly79ins5")
+(def protein-insertion3s "Arg78_Gly79insX[5]")
(def protein-insertion3 (mut/protein-insertion "Arg" (coord/protein-coordinate 78)
"Gly" (coord/protein-coordinate 79)
["Xaa" "Xaa" "Xaa" "Xaa" "Xaa"]))
From b551792360f474ffc6a56421eb44110b9bfcea06 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Tue, 15 Oct 2024 09:24:28 +0900
Subject: [PATCH 4/6] fix: add unknown repeat repairer
---
src/clj_hgvs/repairer.cljc | 15 +++++++++++++++
test/clj_hgvs/core_test.cljc | 5 +++++
2 files changed, 20 insertions(+)
diff --git a/src/clj_hgvs/repairer.cljc b/src/clj_hgvs/repairer.cljc
index ec76d7d..0470446 100644
--- a/src/clj_hgvs/repairer.cljc
+++ b/src/clj_hgvs/repairer.cljc
@@ -156,6 +156,20 @@
(string/replace s #"del\d+ins" "delins")
s))
+;; g.1134_1135ins(100) -> g.1134_1135insN[100]
+;; r.431_432ins(5) -> r.431_432insn[5]
+;; p.R78_G79ins23 -> p.R78_G79insX[23]
+(defn ^:no-doc replace-uncertain-bases
+ [s kind]
+ (case kind
+ (:genome :mitochondria :coding-dna :non-coding-dna :circular-dna)
+ (string/replace s #"ins\((\d+)\)$" "insN[$1]")
+ (:rna)
+ (string/replace s #"ins\((\d+)\)$" "insn[$1]")
+ (:protein)
+ (string/replace s #"ins(\d+)$" "insX[$1]")
+ s))
+
;; c.112GAT(14) -> c.112GAT[14]
(defn ^:no-doc replace-repeated-seqs-parens1
[s kind]
@@ -318,6 +332,7 @@
remove-alternative
remove-inv-bases
remove-del-count-from-delins
+ replace-uncertain-bases
replace-repeated-seqs-parens1
replace-repeated-seqs-parens2
remove-genomic-bases-from-protein
diff --git a/test/clj_hgvs/core_test.cljc b/test/clj_hgvs/core_test.cljc
index e2feeec..84ec891 100644
--- a/test/clj_hgvs/core_test.cljc
+++ b/test/clj_hgvs/core_test.cljc
@@ -332,6 +332,11 @@
;; remove-del-count-from-delins
"c.123_124del2insCTGA" "c.123_124delinsCTGA"
+ ;; replace-uncertain-bases
+ "g.1134_1135ins(100)" "g.1134_1135insN[100]"
+ "r.431_432ins(5)" "r.431_432insn[5]"
+ "p.R78_G79ins23" "p.R78_G79insX[23]"
+
;; replace-repeated-seqs-parens1
"c.112GAT(14)" "c.112GAT[14]"
From 4a0f78802ded0ca8b3b7d19562b633c8aa4664c1 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Wed, 16 Oct 2024 09:57:10 +0900
Subject: [PATCH 5/6] feat: support indel unknown insertion and ins-format
---
src/clj_hgvs/mutation.cljc | 68 ++++++++++++++-----
test/clj_hgvs/mutation_test.cljc | 112 ++++++++++++++++++++++++++-----
2 files changed, 146 insertions(+), 34 deletions(-)
diff --git a/src/clj_hgvs/mutation.cljc b/src/clj_hgvs/mutation.cljc
index 7041543..af01791 100644
--- a/src/clj_hgvs/mutation.cljc
+++ b/src/clj_hgvs/mutation.cljc
@@ -1062,14 +1062,19 @@
(defrecord RNAInsertion [coord-start coord-end alt]
Mutation
(format [this] (format this nil))
- (format [this _]
+ (format [this {:keys [ins-format] :or {ins-format :auto}}]
(str (coord/format coord-start)
"_"
(coord/format coord-end)
"ins"
(cond
(map? alt) (str (:genbank alt) ":" (:coord-start alt) "_" (:coord-end alt))
- (re-matches #"n{2,}" alt) (str "n[" (count alt) "]")
+ (re-matches #"n{2,}" alt) (case ins-format
+ :auto (if (>= (count alt) 10)
+ (str "n[" (count alt) "]")
+ alt)
+ :bases alt
+ :count (str "n[" (count alt) "]"))
:else alt)))
(plain [this]
(into {:mutation "rna-insertion"} (plain-coords this))))
@@ -1238,20 +1243,28 @@
;;; e.g. r.775delinsga
;;; r.775deluinsga
;;; r.775_777delinsc
+;;; r.775_777delinsn[10]
(defrecord RNAIndel [coord-start coord-end ref alt]
Mutation
(format [this] (format this nil))
- (format [this {:keys [show-bases?] :or {show-bases? false}}]
+ (format [this {:keys [show-bases? ins-format] :or {show-bases? false ins-format :auto}}]
(str (coord/format coord-start)
- (if (and coord-end
- (or (not (coord/comparable-coordinates? coord-start coord-end))
- (neg? (compare coord-start coord-end))))
+ (when (and coord-end
+ (or (not (coord/comparable-coordinates? coord-start coord-end))
+ (neg? (compare coord-start coord-end))))
(str "_" (coord/format coord-end)))
"del"
- (if show-bases? ref)
+ (when show-bases? ref)
"ins"
- alt))
+ (if (re-matches #"n{2,}" alt)
+ (case ins-format
+ :auto (if (>= (count alt) 10)
+ (str "n[" (count alt) "]")
+ alt)
+ :bases alt
+ :count (str "n[" (count alt) "]"))
+ alt)))
(plain [this]
(into {:mutation "rna-indel"} (plain-coords this)))
@@ -1287,7 +1300,7 @@
(RNAIndel. coord-start coord-end ref alt))
(def ^:private rna-indel-re
- #"([\d\-\+\*]+)(?:_([\d\-\+\*]+))?del([a-z]+)?ins([a-z]+)")
+ #"([\d\-\+\*]+)(?:_([\d\-\+\*]+))?del([a-z]+)?ins([a-z\d\[\]]+)")
(defn parse-rna-indel
[s]
@@ -1295,7 +1308,9 @@
(rna-indel (coord/parse-rna-coordinate coord-s)
(some-> coord-e coord/parse-rna-coordinate)
ref
- alt)))
+ (if (re-find #"n\[\d+\]" alt)
+ (parse-rna-alt-n alt)
+ alt))))
(defmethod restore "rna-indel"
[m]
@@ -1749,7 +1764,7 @@
(defrecord ProteinInsertion [ref-start coord-start ref-end coord-end alts]
Mutation
(format [this] (format this nil))
- (format [this {:keys [amino-acid-format] :or {amino-acid-format :long}}]
+ (format [this {:keys [amino-acid-format ins-format] :or {amino-acid-format :long ins-format :auto}}]
(apply str (flatten [(cond-> ref-start
(= amino-acid-format :short) ->short-amino-acid)
(coord/format coord-start)
@@ -1759,7 +1774,14 @@
(coord/format coord-end)
"ins"
(if (every? #(= % "Xaa") alts)
- (str "X[" (count alts) "]")
+ (let [alts (cond->> alts
+ (= amino-acid-format :short) (map ->short-amino-acid))]
+ (case ins-format
+ :auto (if (>= (count alts) 10)
+ (str "X[" (count alts) "]")
+ alts)
+ :amino-acids alts
+ :count (str "X[" (count alts) "]")))
(cond->> alts
(= amino-acid-format :short) (map ->short-amino-acid)))])))
(plain [this]
@@ -1817,22 +1839,32 @@
;;;
;;; e.g. Cys28delinsTrpVal
;;; Cys28_Lys29delinsTrp
+;;; Cys28_Lys29delinsX[10]
(defrecord ProteinIndel [ref-start coord-start ref-end coord-end alts]
Mutation
(format [this] (format this nil))
- (format [this {:keys [amino-acid-format] :or {amino-acid-format :long}}]
+ (format [this {:keys [amino-acid-format ins-format] :or {amino-acid-format :long ins-format :auto}}]
(apply str (flatten [(cond-> ref-start
(= amino-acid-format :short) ->short-amino-acid)
(coord/format coord-start)
- (if (should-show-end? ref-start coord-start ref-end coord-end)
+ (when (should-show-end? ref-start coord-start ref-end coord-end)
["_"
(cond-> ref-end
(= amino-acid-format :short) ->short-amino-acid)
(coord/format coord-end)])
"delins"
- (cond->> alts
- (= amino-acid-format :short) (map ->short-amino-acid))])))
+ (if (every? #(= % "Xaa") alts)
+ (let [alts (cond->> alts
+ (= amino-acid-format :short) (map ->short-amino-acid))]
+ (case ins-format
+ :auto (if (>= (count alts) 10)
+ (str "X[" (count alts) "]")
+ alts)
+ :amino-acids alts
+ :count (str "X[" (count alts) "]")))
+ (cond->> alts
+ (= amino-acid-format :short) (map ->short-amino-acid)))])))
(plain [this]
(into {:mutation "protein-indel"} (plain-coords this))))
@@ -1859,7 +1891,7 @@
(ProteinIndel. ref-start coord-start ref-end coord-end alts))
(def ^:private protein-indel-re
- #"([A-Z](?:[a-z]{2})?)(\d+)(?:_([A-Z](?:[a-z]{2})?)(\d+))?delins([A-Z*][a-zA-Z*]*)?")
+ #"([A-Z](?:[a-z]{2})?)(\d+)(?:_([A-Z](?:[a-z]{2})?)(\d+))?delins([A-Z*][a-zA-Z*\[\]\d]*)?")
(defn parse-protein-indel
[s]
@@ -1868,7 +1900,7 @@
(coord/parse-protein-coordinate coord-s)
(->long-amino-acid ref-e)
(some-> coord-e coord/parse-protein-coordinate)
- (mapv ->long-amino-acid (some->> alts (re-seq #"[A-Z*](?:[a-z]{2})?"))))))
+ (mapv ->long-amino-acid (some->> alts parse-protein-insertion-alts)))))
(defmethod restore "protein-indel"
[m]
diff --git a/test/clj_hgvs/mutation_test.cljc b/test/clj_hgvs/mutation_test.cljc
index aa1e52d..85061d7 100644
--- a/test/clj_hgvs/mutation_test.cljc
+++ b/test/clj_hgvs/mutation_test.cljc
@@ -831,13 +831,19 @@
:coord-end (coord/plain (coord/rna-coordinate 757 nil nil))
:alt "acu"})
-(def rna-insertion2s "431_432insn[5]")
+(def rna-insertion2sb "431_432insnnnnn")
+(def rna-insertion2sc "431_432insn[5]")
(def rna-insertion2 (mut/rna-insertion (coord/rna-coordinate 431 nil nil)
(coord/rna-coordinate 432 nil nil)
"nnnnn"))
-(def rna-insertion3s "123_124insL37425.1:23_361")
-(def rna-insertion3 (mut/rna-insertion (coord/rna-coordinate 123 nil nil)
+(def rna-insertion3sc "431_432insn[10]")
+(def rna-insertion3 (mut/rna-insertion (coord/rna-coordinate 431 nil nil)
+ (coord/rna-coordinate 432 nil nil)
+ "nnnnnnnnnn"))
+
+(def rna-insertion4s "123_124insL37425.1:23_361")
+(def rna-insertion4 (mut/rna-insertion (coord/rna-coordinate 123 nil nil)
(coord/rna-coordinate 124 nil nil)
{:genbank "L37425.1"
:coord-start 23
@@ -845,17 +851,22 @@
(deftest format-rna-insertion-test
(testing "returns a string expression of a RNA insertion"
- (are [m s] (= (mut/format m nil) s)
- rna-insertion1 rna-insertion1s
- rna-insertion2 rna-insertion2s
- rna-insertion3 rna-insertion3s)))
+ (are [m o s] (= (mut/format m o) s)
+ rna-insertion1 nil rna-insertion1s
+ rna-insertion2 {:ins-format :auto} rna-insertion2sb
+ rna-insertion2 {:ins-format :bases} rna-insertion2sb
+ rna-insertion2 {:ins-format :count} rna-insertion2sc
+ rna-insertion3 {:ins-format :auto} rna-insertion3sc
+ rna-insertion4 nil rna-insertion4s)))
(deftest parse-rna-insertion-test
(testing "returns a correct RNAInsertion"
(are [s m] (= (mut/parse-rna-insertion s) m)
rna-insertion1s rna-insertion1
- rna-insertion2s rna-insertion2
- rna-insertion3s rna-insertion3))
+ rna-insertion2sb rna-insertion2
+ rna-insertion2sc rna-insertion2
+ rna-insertion3sc rna-insertion3
+ rna-insertion4s rna-insertion4))
(testing "invalid RNA insertion"
(are [s] (thrown? #?(:clj Throwable, :cljs js/Error)
(mut/parse-rna-insertion s))
@@ -969,20 +980,40 @@
nil
"c"))
+(def rna-indel4sb "775_777delinsnnnnn")
+(def rna-indel4sc "775_777delinsn[5]")
+(def rna-indel4 (mut/rna-indel (coord/rna-coordinate 775 nil nil)
+ (coord/rna-coordinate 777 nil nil)
+ nil
+ "nnnnn"))
+
+(def rna-indel5sc "775_777delinsn[10]")
+(def rna-indel5 (mut/rna-indel (coord/rna-coordinate 775 nil nil)
+ (coord/rna-coordinate 777 nil nil)
+ nil
+ "nnnnnnnnnn"))
+
(deftest format-rna-indel-test
(testing "returns a string expression of a RNA indel"
(are [m o s] (= (mut/format m o) s)
rna-indel1 nil rna-indel1s
rna-indel2 nil rna-indel2ss
rna-indel2 {:show-bases? true} rna-indel2s
- rna-indel3 nil rna-indel3s)))
+ rna-indel3 nil rna-indel3s
+ rna-indel4 {:ins-format :auto} rna-indel4sb
+ rna-indel4 {:ins-format :bases} rna-indel4sb
+ rna-indel4 {:ins-format :count} rna-indel4sc
+ rna-indel5 {:ins-format :auto} rna-indel5sc)))
(deftest parse-rna-indel-test
(testing "returns a correct RNAIndel"
(are [s m] (= (mut/parse-rna-indel s) m)
rna-indel1s rna-indel1
rna-indel2s rna-indel2
- rna-indel3s rna-indel3))
+ rna-indel3s rna-indel3
+ rna-indel4sb rna-indel4
+ rna-indel4sc rna-indel4
+ rna-indel5sc rna-indel5))
(testing "invalid RNA indel"
(are [s] (thrown? #?(:clj Throwable, :cljs js/Error)
(mut/parse-rna-indel s))
@@ -1327,11 +1358,20 @@
"Leu" (coord/protein-coordinate 24)
["Arg" "Ser" "Ter"]))
-(def protein-insertion3s "Arg78_Gly79insX[5]")
+(def protein-insertion3sal "Arg78_Gly79insXaaXaaXaaXaaXaa")
+(def protein-insertion3sas "R78_G79insXXXXX")
+(def protein-insertion3sc "Arg78_Gly79insX[5]")
+(def protein-insertion3scs "R78_G79insX[5]")
(def protein-insertion3 (mut/protein-insertion "Arg" (coord/protein-coordinate 78)
"Gly" (coord/protein-coordinate 79)
["Xaa" "Xaa" "Xaa" "Xaa" "Xaa"]))
+(def protein-insertion4sc "Arg78_Gly79insX[10]")
+(def protein-insertion4scs "R78_G79insX[10]")
+(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"]))
+
(deftest format-protein-insertion-test
(testing "returns a string expression of a protein insertion"
(are [m o s] (= (mut/format m o) s)
@@ -1339,7 +1379,14 @@
protein-insertion1 {:amino-acid-format :short} protein-insertion1ss
protein-insertion2 nil protein-insertion2s
protein-insertion2 {:amino-acid-format :short} protein-insertion2ss
- protein-insertion3 nil protein-insertion3s)))
+ protein-insertion3 {:ins-format :auto} protein-insertion3sal
+ protein-insertion3 {:amino-acid-format :short :ins-format :auto} protein-insertion3sas
+ protein-insertion3 {:ins-format :amino-acids} protein-insertion3sal
+ protein-insertion3 {:amino-acid-format :short :ins-format :amino-acids} protein-insertion3sas
+ 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)))
(deftest parse-protein-insertion-test
(testing "returns a correct ProteinInsertion"
@@ -1348,7 +1395,12 @@
protein-insertion1ss protein-insertion1
protein-insertion2s protein-insertion2
protein-insertion2ss protein-insertion2
- protein-insertion3s protein-insertion3))
+ protein-insertion3sal protein-insertion3
+ protein-insertion3sas protein-insertion3
+ protein-insertion3sc protein-insertion3
+ protein-insertion3scs protein-insertion3
+ protein-insertion4sc protein-insertion4
+ protein-insertion4scs protein-insertion4))
(testing "invalid protein insertion"
(are [s] (thrown? #?(:clj Throwable, :cljs js/Error)
(mut/parse-protein-insertion s))
@@ -1389,6 +1441,20 @@
"Lys" (coord/protein-coordinate 29)
["Ter"]))
+(def protein-indel4sal "Cys28_Lys29delinsXaaXaaXaaXaaXaa")
+(def protein-indel4sas "C28_K29delinsXXXXX")
+(def protein-indel4sc "Cys28_Lys29delinsX[5]")
+(def protein-indel4scs "C28_K29delinsX[5]")
+(def protein-indel4 (mut/protein-indel "Cys" (coord/protein-coordinate 28)
+ "Lys" (coord/protein-coordinate 29)
+ ["Xaa" "Xaa" "Xaa" "Xaa" "Xaa"]))
+
+(def protein-indel5sc "Cys28_Lys29delinsX[10]")
+(def protein-indel5scs "C28_K29delinsX[10]")
+(def protein-indel5 (mut/protein-indel "Cys" (coord/protein-coordinate 28)
+ "Lys" (coord/protein-coordinate 29)
+ ["Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa" "Xaa"]))
+
(deftest format-protein-indel-test
(testing "returns a string expression of a protein indel"
(are [m o s] (= (mut/format m o) s)
@@ -1397,7 +1463,15 @@
protein-indel2 nil protein-indel2s
protein-indel2 {:amino-acid-format :short} protein-indel2ss
protein-indel3 nil protein-indel3s
- protein-indel3 {:amino-acid-format :short} protein-indel3ss)))
+ protein-indel3 {:amino-acid-format :short} protein-indel3ss
+ protein-indel4 {:ins-format :auto} protein-indel4sal
+ protein-indel4 {:amino-acid-format :short :ins-format :auto} protein-indel4sas
+ protein-indel4 {:ins-format :amino-acids} protein-indel4sal
+ protein-indel4 {:amino-acid-format :short :ins-format :amino-acids} protein-indel4sas
+ protein-indel4 {:ins-format :count} protein-indel4sc
+ protein-indel4 {:amino-acid-format :short :ins-format :count} protein-indel4scs
+ protein-indel5 {:ins-format :auto} protein-indel5sc
+ protein-indel5 {:amino-acid-format :short :ins-format :auto} protein-indel5scs)))
(deftest parse-protein-indel-test
(testing "returns a correct ProteinIndel"
@@ -1407,7 +1481,13 @@
protein-indel2s protein-indel2
protein-indel2ss protein-indel2
protein-indel3s protein-indel3
- protein-indel3ss protein-indel3))
+ protein-indel3ss protein-indel3
+ protein-indel4sal protein-indel4
+ protein-indel4sas protein-indel4
+ protein-indel4sc protein-indel4
+ protein-indel4scs protein-indel4
+ protein-indel5sc protein-indel5
+ protein-indel5scs protein-indel5))
(testing "invalid protein indel"
(are [s] (thrown? #?(:clj Throwable, :cljs js/Error)
(mut/parse-protein-indel s))
From 36d828df711d6e71ecf600853c8d8e35faaa2ab9 Mon Sep 17 00:00:00 2001
From: Nobuaki Karasawa <nkarasawa@xcoo.jp>
Date: Wed, 16 Oct 2024 10:10:52 +0900
Subject: [PATCH 6/6] chore: Prepare for next development iteration
(0.5.0-SNAPSHOT)
---
project.clj | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/project.clj b/project.clj
index 0af15e3..d0d6315 100644
--- a/project.clj
+++ b/project.clj
@@ -1,4 +1,4 @@
-(defproject clj-hgvs "0.4.7"
+(defproject clj-hgvs "0.5.0-SNAPSHOT"
:description "Clojure(Script) library for handling HGVS"
:url "https://github.com/chrovis/clj-hgvs"
:license {:name "Apache License, Version 2.0"