CLN-32292: update aminoAcidToDegenerateDnaMap.js

src/aminoAcidToDegenerateDnaMap.js

@@ -3,28 +3,30 @@ const aminoAcidToDegenerateDnaMap = {
   ".": "...",
   "*": "trr",
   a: "gcn",
-  b: "ray",
-  c: "ugy",
+  b: "ray", // D or N => aay + gay = ray
+  c: "tgy",
   d: "gay",
   e: "gar",
-  f: "uuy",
+  f: "tty",
   g: "ggn",
   h: "cay",
-  i: "auh",
+  i: "ath",
+  j: "htn", // L or I ytn + ath => htn
   k: "aar",
-  l: "yun",
-  m: "aug",
+  l: "ytn", // YTR、CTN => Y=CT, N=AGCT
+  m: "atg",
   n: "aay",
+  o: "tag", // Pyrrolysine, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2933860/
   p: "ccn",
   q: "car",
-  r: "mng",
-  s: "wsn",
+  r: "mgn", // CGN、MGR => M=AC, N=AGCT
+  s: "wsn", // TCN、AGY => AT = w, CG = S, N is AGCT
   t: "acn",
-  u: "uga",
-  v: "gun",
-  w: "ugg",
-  x: "nnn",
-  y: "uay",
-  z: "sar"
+  u: "tga", // Selenocysteine, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2933860/
+  v: "gtn",
+  w: "tgg",
+  x: "nnn", // unknown aa.
+  y: "tay",
+  z: "sar" // E or Q, => gar + car = sar
 };
 module.exports = aminoAcidToDegenerateDnaMap;

src/aminoAcidToDegenerateRnaMap.js

@@ -0,0 +1,32 @@
+const aminoAcidToDegenerateRnaMap = {
+  "-": "---",
+  ".": "...",
+  "*": "trr",
+  a: "gcn",
+  b: "ray",
+  c: "ugy",
+  d: "gay",
+  e: "gar",
+  f: "uuy",
+  g: "ggn",
+  h: "cay",
+  i: "auh",
+  j: "hun",
+  k: "aar",
+  l: "yun",
+  m: "aug",
+  n: "aay",
+  o: "uag",
+  p: "ccn",
+  q: "car",
+  r: "mgn",
+  s: "wsn",
+  t: "acn",
+  u: "uga",
+  v: "gun",
+  w: "ugg",
+  x: "nnn",
+  y: "uay",
+  z: "sar"
+};
+module.exports = aminoAcidToDegenerateRnaMap;

src/aminoAcidToDnaRna.test.js

@@ -0,0 +1,27 @@
+const { expect } = require("chai");
+const getDegenerateDnaStringFromAAString = require("./getDegenerateDnaStringFromAAString");
+const getDegenerateRnaStringFromAAString = require("./getDegenerateRnaStringFromAAString");
+
+describe("amino acid to RNA or DNA should be correct", function() {
+  it('should return a string with no "u/U" in it when parse AA sequence to DNA sequence.', function() {
+    const aaStr = "AQRSTFFVCL";
+    const DNASeq = getDegenerateDnaStringFromAAString(aaStr);
+    const RNASeq = getDegenerateRnaStringFromAAString(aaStr);
+    expect(DNASeq.length)
+      .equal(RNASeq.length)
+      .equal(aaStr.length * 3);
+    expect(DNASeq.toLowerCase().includes("u")).equal(false);
+    expect(DNASeq.toLowerCase().includes("t")).equal(true);
+
+    expect(RNASeq.toLowerCase().includes("t")).equal(false);
+    expect(RNASeq.toLowerCase().includes("u")).equal(true);
+
+    expect(RNASeq.toLowerCase().replace(/u/gi, "t")).equal(
+      DNASeq.toLowerCase()
+    );
+
+    expect(DNASeq.toLowerCase().replace(/t/gi, "u")).equal(
+      RNASeq.toLowerCase()
+    );
+  });
+});

src/degenerateRnaToAminoAcidMap.js

@@ -0,0 +1,5 @@
+const { invert } = require("lodash");
+const aminoAcidToDegenerateRnaMap = require("./aminoAcidToDegenerateRnaMap");
+
+const degenerateRnaToAminoAcidMap = invert(aminoAcidToDegenerateRnaMap);
+module.exports = degenerateRnaToAminoAcidMap;

src/getDegenerateRnaStringFromAAString.js

@@ -0,0 +1,8 @@
+const aminoAcidToDegenerateRnaMap = require("./aminoAcidToDegenerateRnaMap");
+
+module.exports = function getDegenerateRnaStringFromAAString(aaString) {
+  return aaString
+    .split("")
+    .map(char => aminoAcidToDegenerateRnaMap[char.toLowerCase()] || "nnn")
+    .join("");
+};

src/index.js

@@ -9,6 +9,7 @@ module.exports.patchSeqWithDiff = diffUtils.patchSeqWithDiff;
 module.exports.reverseSeqDiff = diffUtils.reverseSeqDiff;
 module.exports.getAllInsertionsInSeqReads = require("./getAllInsertionsInSeqReads");
 module.exports.getDegenerateDnaStringFromAAString = require("./getDegenerateDnaStringFromAAString");
+module.exports.getDegenerateRnaStringFromAAString = require("./getDegenerateRnaStringFromAAString");
 module.exports.getVirtualDigest = require("./getVirtualDigest");
 module.exports.isEnzymeType2S = require("./isEnzymeType2S");
 module.exports.insertGapsIntoRefSeq = require("./insertGapsIntoRefSeq");
@@ -72,4 +73,6 @@ module.exports.getDigestFragmentsForRestrictionEnzymes = require("./getDigestFra
 module.exports.convertDnaCaretPositionOrRangeToAA = require("./convertDnaCaretPositionOrRangeToAA");
 module.exports.convertAACaretPositionOrRangeToDna = require("./convertAACaretPositionOrRangeToDna");
 module.exports.aminoAcidToDegenerateDnaMap = require("./aminoAcidToDegenerateDnaMap");
+module.exports.aminoAcidToDegenerateRnaMap = require("./aminoAcidToDegenerateRnaMap");
 module.exports.degenerateDnaToAminoAcidMap = require("./degenerateDnaToAminoAcidMap");
+module.exports.degenerateRnaToAminoAcidMap = require("./degenerateRnaToAminoAcidMap");

src/insertSequenceDataAtPositionOrRange.test.js

@@ -34,7 +34,7 @@ describe("insertSequenceData", function() {
       sequenceToInsertInto,
       caret
     );
-    postInsertSeq.sequence.should.equal("atgauhgaymngagagagaaa");
+    postInsertSeq.sequence.should.equal("atgathgaymgnagagagaaa");
     postInsertSeq.proteinSequence.should.equal("MIDRREK");
     postInsertSeq.features.should.containSubset([
       { name: "feat2", start: 0, end: 14 }

src/tidyUpSequenceData.test.js

@@ -126,7 +126,7 @@ describe("tidyUpSequenceData", function() {
       isProtein: true,
       size: 54, //size should refer to the DNA length
       proteinSize: 18, //proteinSize should refer to the amino acid length
-      sequence: "ggngcnggnauhugacayuggggngcnggngcnyungcnwsnggnyuntrr...", //degenerate sequence
+      sequence: "ggngcnggnathtgacaytggggngcnggngcnytngcnwsnggnytntrr...", //degenerate sequence
       proteinSequence: "gagiuhwgagalasgl*.",
       circular: false,
       features: [