@teselagen/bio-parsers 0.4.15 → 0.4.17

package/index.mjs

@@ -11504,6 +11504,26 @@ function getDegenerateDnaStringFromAAString(aaString) {
   return aaString.split("").map((char) => aminoAcidToDegenerateDnaMap[char.toLowerCase()] || "nnn").join("");
 }
 __name(getDegenerateDnaStringFromAAString, "getDegenerateDnaStringFromAAString");
+function getAminoAcidStringFromSequenceString(sequenceString, { doNotExcludeAsterisk } = {}) {
+  const aminoAcidsPerBase = getAminoAcidDataForEachBaseOfDna(
+    sequenceString,
+    true
+  );
+  const aaArray = [];
+  let aaString = "";
+  aminoAcidsPerBase.forEach((aa, index) => {
+    if (!aa.fullCodon) {
+      return;
+    }
+    if (!doNotExcludeAsterisk && index >= aminoAcidsPerBase.length - 3 && aa.aminoAcid.value === "*") {
+      return;
+    }
+    aaArray[aa.aminoAcidIndex] = aa.aminoAcid.value;
+  });
+  aaString = aaArray.join("");
+  return aaString;
+}
+__name(getAminoAcidStringFromSequenceString, "getAminoAcidStringFromSequenceString");
 function tidyUpSequenceData(pSeqData, options = {}) {
   const {
     annotationsAsObjects,
@@ -11511,6 +11531,7 @@ function tidyUpSequenceData(pSeqData, options = {}) {
     doNotRemoveInvalidChars,
     additionalValidChars,
     noTranslationData,
+    includeProteinSequence,
     doNotProvideIdsForAnnotations,
     noCdsTranslations,
     convertAnnotationsFromAAIndices,
@@ -11544,7 +11565,9 @@ function tidyUpSequenceData(pSeqData, options = {}) {
   }
   if (!doNotRemoveInvalidChars) {
     if (seqData.isProtein) {
-      const [newSeq] = filterSequenceString(seqData.proteinSequence, __spreadValues({}, topLevelSeqData || seqData));
+      const [newSeq] = filterSequenceString(seqData.proteinSequence, __spreadProps(__spreadValues({}, topLevelSeqData || seqData), {
+        isProtein: true
+      }));
       seqData.proteinSequence = newSeq;
     } else {
       const [newSeq] = filterSequenceString(seqData.sequence, __spreadValues({
@@ -11565,6 +11588,10 @@ function tidyUpSequenceData(pSeqData, options = {}) {
       null,
       true
     );
+  } else if (includeProteinSequence) {
+    seqData.proteinSequence = getAminoAcidStringFromSequenceString(
+      seqData.sequence
+    );
   }
   seqData.size = seqData.noSequence ? seqData.size : seqData.sequence.length;
   seqData.proteinSize = seqData.noSequence ? seqData.proteinSize : seqData.proteinSequence.length;
@@ -19785,19 +19812,38 @@ function flattenSequenceArray(parsingResultArray, opts) {
   return parsingResultArray;
 }
 __name(flattenSequenceArray, "flattenSequenceArray");
+function wrapOriginSpanningFeatures(locArrayInput, sequenceLength, inclusive1BasedStart, inclusive1BasedEnd) {
+  const locArrayOutput = locArrayInput.map((loc) => __spreadValues({}, loc));
+  for (let i = 0; i < locArrayOutput.length - 1; i++) {
+    const firstFeature = locArrayOutput[i];
+    const secondFeature = locArrayOutput[i + 1];
+    if (firstFeature.end === sequenceLength - (inclusive1BasedEnd ? 0 : 1) && secondFeature.start === 1 - (inclusive1BasedStart ? 0 : 1)) {
+      locArrayOutput[i] = {
+        start: firstFeature.start,
+        end: secondFeature.end
+      };
+      locArrayOutput.splice(i + 1, 1);
+    }
+  }
+  return locArrayOutput;
+}
+__name(wrapOriginSpanningFeatures, "wrapOriginSpanningFeatures");
 function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive1BasedEnd, isCircular, sequenceLength) {
   locStr = locStr.trim();
-  const locArr = [];
-  locStr.replace(/(\d+)/g, function(string, match) {
-    locArr.push(match);
+  const positionsArray = [];
+  const locationParts = locStr.split(",");
+  locationParts.forEach((locPart) => {
+    const extractedPositions = locPart.match(/(\d+)/g);
+    if (extractedPositions === null) {
+      return;
+    }
+    positionsArray.push(extractedPositions[0]);
+    positionsArray.push(extractedPositions[1] || extractedPositions[0]);
   });
   const locArray = [];
-  for (let i = 0; i < locArr.length; i += 2) {
-    const start = parseInt(locArr[i], 10) - (inclusive1BasedStart ? 0 : 1);
-    let end = parseInt(locArr[i + 1], 10) - (inclusive1BasedEnd ? 0 : 1);
-    if (isNaN(end)) {
-      end = start;
-    }
+  for (let i = 0; i < positionsArray.length; i += 2) {
+    const start = parseInt(positionsArray[i], 10) - (inclusive1BasedStart ? 0 : 1);
+    const end = parseInt(positionsArray[i + 1], 10) - (inclusive1BasedEnd ? 0 : 1);
     const location = {
       start,
       end
@@ -19806,20 +19852,16 @@ function parseFeatureLocation(locStr, isProtein, inclusive1BasedStart, inclusive
       isProtein ? convertAACaretPositionOrRangeToDna(location) : location
     );
   }
-  if (isCircular) {
-    for (let i = 0; i < locArray.length; i += 2) {
-      const firstFeature = locArray[i];
-      const secondFeature = locArray[i + 1];
-      if (firstFeature.end === sequenceLength - (inclusive1BasedEnd ? 0 : 1) && secondFeature.start === 1 - (inclusive1BasedStart ? 0 : 1)) {
-        locArray[i] = {
-          start: firstFeature.start,
-          end: secondFeature.end
-        };
-        locArray.splice(i + 1, 1);
-      }
-    }
+  if (isCircular && sequenceLength) {
+    return wrapOriginSpanningFeatures(
+      locArray,
+      sequenceLength,
+      inclusive1BasedStart,
+      inclusive1BasedEnd
+    );
+  } else {
+    return locArray;
   }
-  return locArray;
 }
 __name(parseFeatureLocation, "parseFeatureLocation");
 function genbankToJson(string, options = {}) {
@@ -19916,7 +19958,7 @@ function genbankToJson(string, options = {}) {
           parseOrigin(line, key);
           break;
         case genbankAnnotationKey.END_SEQUENCE_TAG:
-          endSeq();
+          endSeq(options);
           break;
         case genbankAnnotationKey.DEFINITION_TAG:
           line = line.replace(/DEFINITION/, "");
@@ -20021,9 +20063,9 @@ function genbankToJson(string, options = {}) {
     }
   });
   return results;
-  function endSeq() {
+  function endSeq(options2) {
     hasFoundLocus = false;
-    postProcessCurSeq();
+    postProcessCurSeq(options2);
     resultsArray.push(result || { success: false });
   }
   __name(endSeq, "endSeq");
@@ -20037,11 +20079,13 @@ function genbankToJson(string, options = {}) {
     }
   }
   __name(addMessage, "addMessage");
-  function postProcessCurSeq() {
+  function postProcessCurSeq(options2) {
     if (result && result.parsedSequence && result.parsedSequence.features) {
       for (let i = 0; i < result.parsedSequence.features.length; i++) {
         result.parsedSequence.features[i] = postProcessGenbankFeature(
-          result.parsedSequence.features[i]
+          result.parsedSequence.features[i],
+          result.parsedSequence,
+          options2
         );
       }
     }
@@ -20287,7 +20331,7 @@ function genbankToJson(string, options = {}) {
     return runon;
   }
   __name(isKeywordRunon, "isKeywordRunon");
-  function postProcessGenbankFeature(feat) {
+  function postProcessGenbankFeature(feat, parsedSequence, options2) {
     if (feat.notes.label) {
       feat.name = feat.notes.label[0];
     } else if (feat.notes.gene) {
@@ -20320,6 +20364,15 @@ function genbankToJson(string, options = {}) {
       feat.arrowheadType = feat.notes.direction[0].toUpperCase() === "BOTH" ? "BOTH" : feat.notes.direction[0].toUpperCase() === "NONE" ? "NONE" : void 0;
       delete feat.notes.direction;
     }
+    if (parsedSequence.circular) {
+      const { inclusive1BasedStart: inclusive1BasedStart2, inclusive1BasedEnd: inclusive1BasedEnd2 } = options2;
+      feat.locations = wrapOriginSpanningFeatures(
+        feat.locations,
+        parsedSequence.sequence.length,
+        inclusive1BasedStart2,
+        inclusive1BasedEnd2
+      );
+    }
     return feat;
   }
   __name(postProcessGenbankFeature, "postProcessGenbankFeature");