support RNA sequence in gb file

packages/bio-parsers/src/genbankToJson.js

@@ -338,7 +338,7 @@ function genbankToJson(string, options = {}) {
 
       if (
         j === 4 &&
-        (item.match(/ds-dna/i) || item.match(/ss-dna/i) || item.match(/dna/i))
+        (item.match(/ds-dna/i) || item.match(/ss-dna/i) || item.match(/dna/i) || item.match(/rna/i))
       ) {
         if (options.isProtein === undefined) {
           options.isProtein = false;
@@ -347,6 +347,9 @@ function genbankToJson(string, options = {}) {
         if (item.match(/ss-dna/i)) {
           options.isSingleStrandedDNA = true;
         }
+        if (item.match(/rna/i) && !item.match(/ss-rna/i)) {
+          options.isDoubleStrandedRNA = true;
+        }
       }
 
       // Division
@@ -368,6 +371,7 @@ function genbankToJson(string, options = {}) {
     result.parsedSequence.gbDivision = gbDivision;
     result.parsedSequence.sequenceTypeFromLocus = options.sequenceTypeFromLocus;
     result.parsedSequence.isSingleStrandedDNA = options.isSingleStrandedDNA;
+    result.parsedSequence.isDoubleStrandedRNA = options.isDoubleStrandedRNA;
     result.parsedSequence.date = date;
     result.parsedSequence.circular = circular;
   }

packages/bio-parsers/src/jsonToGenbank.js

@@ -205,9 +205,9 @@ function createGenbankLocus(serSeq, options) {
   if (serSeq.isProtein) {
     dnaType = "";
   } else if (serSeq.type === "RNA") {
-    dnaType = "RNA";
+    dnaType = serSeq?.doubleStranded ? 'RNA' : serSeq?.sequenceTypeFromLocus ?? "ss-RNA";
   } else {
-    dnaType = serSeq?.sequenceTypeFromLocus ?? "DNA";
+    dnaType = serSeq?.doubleStranded ? 'DNA' : serSeq?.sequenceTypeFromLocus ?? "DNA";
   }
   const date = getCurrentDateString();
 

packages/bio-parsers/test/genbankToJson.test.js

@@ -58,6 +58,70 @@ ORIGIN
       },
     ]);
   });
+
+  it(`correctly handles the single-stranded/double-stranded RNA/DNA in LOCUS line`, () => {
+    const ss_DNA_string = `LOCUS       Tt2-PstI-SphI-rev(dna)        20 bp    ss-DNA     circular
+    04-FEB-2021
+DEFINITION  [Heavy] lalalal
+            more description here
+            and still more
+ACCESSION   Tt2-PstI-SphI-rev
+VERSION     Tt2-PstI-SphI-rev.0
+KEYWORDS    .
+SOURCE      Homo sapiens
+ORGANISM  Homo sapiens
+    .
+COMMENT     Chain:Heavy
+    Numbering:Kabat
+    AnnotationCategory:VREGION
+    Plasmid: pAETEST
+    ClonedAnnotationCategory:VREGION
+ORIGIN
+1 tcgcgcgttt cggtgatgac
+//`;
+
+  const ds_DNA_string = ss_DNA_string.replace('ss-DNA', 'DNA');
+
+  const ss_RNA_string = `LOCUS       Tt2-PstI-SphI-rev(rna)        20 bp    ss-RNA     circular
+    04-FEB-2021
+DEFINITION  [Heavy] lalalal
+            more description here
+            and still more
+ACCESSION   Tt2-PstI-SphI-rev
+VERSION     Tt2-PstI-SphI-rev.0
+KEYWORDS    .
+SOURCE      Homo sapiens
+ORGANISM  Homo sapiens
+    .
+COMMENT     Chain:Heavy
+    Numbering:Kabat
+    AnnotationCategory:VREGION
+    Plasmid: pAETEST
+    ClonedAnnotationCategory:VREGION
+ORIGIN
+1 ucgcgcguuu cggugaugac
+//`;
+
+  const ds_RNA_string = ss_RNA_string.replace('ss-RNA', 'RNA');
+
+
+  const ss_DNA_result = genbankToJson(ss_DNA_string);
+  ss_DNA_result[0].parsedSequence.isSingleStrandedDNA.should.equal(true);
+
+
+  const ds_DNA_result = genbankToJson(ds_DNA_string);
+  Boolean(ds_DNA_result[0].parsedSequence.isSingleStrandedDNA).should.equal(false);
+
+  const ss_RNA_result = genbankToJson(ss_RNA_string);
+  Boolean(ss_RNA_result[0].parsedSequence.isDoubleStrandedRNA).should.equal(false);
+
+
+  const ds_RNA_result = genbankToJson(ds_RNA_string);
+  ds_RNA_result[0].parsedSequence.isDoubleStrandedRNA.should.equal(true);
+});
+
+
+
   it(`correctly handles a multi-line DEFINITION converting it to description`, () => {
     const string = `LOCUS       Tt2-PstI-SphI-rev(dna)        7628 bp    DNA     circular
     04-FEB-2021

packages/bio-parsers/test/jsonToGenbank.test.js

@@ -39,6 +39,54 @@ describe("genbank exporter/parser conversion", function () {
     result[0].parsedSequence.features[0].start.should.equal(3);
     result[0].parsedSequence.features[0].end.should.equal(29);
   });
+
+  it(`should has ss-RNA/RNA/ss-DNA/DNA in the LOCUS line`, () => {
+    const dnaSequence = 'agctttgggttt';
+    const rnaSequence = 'agcuuuggguuu';
+    const dsDNAString1 = jsonToGenbank({
+      sequence: dnaSequence,
+      type: 'DNA'
+    });
+    const dsDNAString2 = jsonToGenbank({
+      sequence: dnaSequence,
+      doubleStranded: true,
+      type: 'DNA'
+    });
+    const ssDNAString = jsonToGenbank({
+      sequence: dnaSequence,
+      sequenceTypeFromLocus: 'ss-DNA',
+      type: 'DNA'
+    });
+
+    assert(dsDNAString1.indexOf('DNA') !== -1);
+    assert(dsDNAString1.indexOf('ss-DNA') === -1);
+    assert(dsDNAString2.indexOf('DNA') !== -1);
+    assert(dsDNAString2.indexOf('ss-DNA') === -1);
+    assert(ssDNAString.indexOf('ss-DNA') !== -1);
+
+    const dsRNAString1 = jsonToGenbank({
+      sequence: rnaSequence,
+      doubleStranded: true,
+      type: 'RNA'
+    });
+    const dsRNAString2 = jsonToGenbank({
+      sequence: rnaSequence,
+      sequenceTypeFromLocus: 'RNA',
+      type: 'RNA'
+    });
+    const ssRNAString = jsonToGenbank({
+      sequence: rnaSequence,
+      sequenceTypeFromLocus: 'ss-RNA',
+      type: 'RNA'
+    });
+
+    assert(dsRNAString1.indexOf('RNA') !== -1);
+    assert(dsRNAString1.indexOf('ss-RNA') === -1);
+    assert(dsRNAString2.indexOf('RNA') !== -1);
+    assert(dsRNAString2.indexOf('ss-RNA') === -1);
+    assert(ssRNAString.indexOf('ss-RNA') !== -1);
+  });
+
   it(`should have a space at the 68 position in the genbank locus `, () => {
     const string = jsonToGenbank({
       sequence: "agagagagagag",

packages/sequence-utils/src/getComplementSequenceAndAnnotations.js

@@ -14,7 +14,7 @@ export default function getComplementSequenceAndAnnotations(
     options
   );
   const newSeqObj = Object.assign({}, seqObj, {
-    sequence: getComplementSequenceString(seqObj.sequence)
+    sequence: getComplementSequenceString(seqObj.sequence, seqObj.isRna)
   });
   return tidyUpSequenceData(newSeqObj, options);
 };

packages/sequence-utils/src/getComplementSequenceString.js

@@ -1,12 +1,14 @@
 import DNAComplementMap from "./DNAComplementMap";
+import { merge } from "lodash";
 
 
 // ac.throw([ac.string,ac.bool],arguments);
-export default function getComplementSequenceString(sequence) {
+export default function getComplementSequenceString(sequence, isRna) {
   // ac.throw([ac.string],arguments);
   let complementSeqString = "";
+  const complementMap = merge(DNAComplementMap, isRna ? { a: 'u', A: 'U'} : {a: 't', A: 'T'});
   for (let i = 0; i < sequence.length; i++) {
-    let complementChar = DNAComplementMap[sequence[i]];
+    let complementChar = complementMap[sequence[i]];
     if (!complementChar) {
       complementChar = sequence[i];
       // throw new Error('trying to get the reverse compelement of an invalid base');

packages/sequence-utils/src/getComplementSequenceString.test.js

@@ -0,0 +1,13 @@
+import getComplementSequenceString from "./getComplementSequenceString";
+
+import assert from "assert";
+
+describe('complement base should be shown correctly', () => {
+  it('complement base should be shown correctly for RNA sequence', () => {
+    assert.equal('UUA', getComplementSequenceString('AAU', true))
+  });
+
+  it('complement base should be shown correctly for DNA sequence', () => {
+    assert.equal('TTA', getComplementSequenceString('AAT'))
+  });
+});