@teselagen/sequence-utils 0.3.10 → 0.3.12

package/bioData.d.ts

@@ -1,6 +1,6 @@
 export const protein_letters: "ACDEFGHIKLMNPQRSTVWY";
 export const protein_letters_withUandX: "ACDEFGHIKLMNPQRSTVWYUX";
-export const extended_protein_letters: "ACDEFGHIKLMNPQRSTVWYBXZJUO*";
+export const extended_protein_letters: "ACDEFGHIKLMNPQRSTVWYBXZJUO";
 export const ambiguous_dna_letters: "GATCRYWSMKHBVDN";
 export const unambiguous_dna_letters: "GATC";
 export const ambiguous_rna_letters: "GAUCRYWSMKHBVDN";

package/filterSequenceString.d.ts

@@ -1,4 +1,4 @@
-export default function filterSequenceString(sequenceString: any, { additionalValidChars, isOligo, name, isProtein, isRna, isMixedRnaAndDna, }?: {
+export default function filterSequenceString(sequenceString: any, { additionalValidChars, isOligo, name, isProtein, isRna, isMixedRnaAndDna }?: {
     additionalValidChars?: string | undefined;
     isOligo: any;
     name: any;
@@ -6,7 +6,7 @@ export default function filterSequenceString(sequenceString: any, { additionalVa
     isRna: any;
     isMixedRnaAndDna: any;
 }): (string | string[])[];
-export function getAcceptedChars({ isOligo, isProtein, isRna, isMixedRnaAndDna, }?: {
+export function getAcceptedChars({ isOligo, isProtein, isRna, isMixedRnaAndDna }?: {
     isOligo: any;
     isProtein: any;
     isRna: any;

package/index.js

@@ -6000,7 +6000,7 @@ lodash.exports;
 var lodashExports = lodash.exports;
 const protein_letters = "ACDEFGHIKLMNPQRSTVWY";
 const protein_letters_withUandX = "ACDEFGHIKLMNPQRSTVWYUX";
-const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO*";
+const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO";
 const ambiguous_dna_letters = "GATCRYWSMKHBVDN";
 const unambiguous_dna_letters = "GATC";
 const ambiguous_rna_letters = "GAUCRYWSMKHBVDN";
@@ -6056,7 +6056,7 @@ const extended_protein_values = {
   Y: "Y",
   Z: "QE",
   "*": "\\*\\.",
-  ".": "\\.\\.",
+  ".": "\\.",
   "-": "\\-"
 };
 const bioData = /* @__PURE__ */ Object.freeze(/* @__PURE__ */ Object.defineProperty({
@@ -12399,9 +12399,12 @@ function getReplaceChars({
   isRna,
   isMixedRnaAndDna
 } = {}) {
-  return isProtein ? {} : isOligo ? {} : isRna ? { t: "u" } : isMixedRnaAndDna ? {} : (
-    //just plain old dna
-    {}
+  return isProtein ? {} : (
+    // {".": "*"}
+    isOligo ? {} : isRna ? { t: "u" } : isMixedRnaAndDna ? {} : (
+      //just plain old dna
+      {}
+    )
   );
 }
 __name(getReplaceChars, "getReplaceChars");
@@ -12672,11 +12675,13 @@ __name(tidyUpSequenceData, "tidyUpSequenceData");
 const getDiffFromSeqs = /* @__PURE__ */ __name((oldData, newData, { ignoreKeys = [] } = {}) => {
   oldData = tidyUpSequenceData(oldData, {
     annotationsAsObjects: true,
-    noTranslationData: true
+    noTranslationData: true,
+    doNotRemoveInvalidChars: true
   });
   newData = tidyUpSequenceData(newData, {
     annotationsAsObjects: true,
-    noTranslationData: true
+    noTranslationData: true,
+    doNotRemoveInvalidChars: true
   });
   [oldData, newData].forEach((d) => {
     [
@@ -12708,7 +12713,10 @@ const patchSeqWithDiff = /* @__PURE__ */ __name((oldData, diff, { ignoreKeys = [
     delete diff[k];
   });
   return jsondiffpatch_umdExports.patch(
-    tidyUpSequenceData(lodashExports.cloneDeep(oldData), { annotationsAsObjects: true }),
+    tidyUpSequenceData(lodashExports.cloneDeep(oldData), {
+      annotationsAsObjects: true,
+      doNotRemoveInvalidChars: true
+    }),
     diff
   );
 }, "patchSeqWithDiff");
@@ -13281,7 +13289,9 @@ function arrayRotate(arr, count) {
 }
 __name(arrayRotate, "arrayRotate");
 function rotateSequenceDataToPosition(sequenceData, caretPosition, options) {
-  const newSequenceData = tidyUpSequenceData(sequenceData, options);
+  const newSequenceData = tidyUpSequenceData(sequenceData, __spreadValues({
+    doNotRemoveInvalidChars: true
+  }, options));
   newSequenceData.sequence = rotateBpsToPosition(
     newSequenceData.sequence,
     caretPosition
@@ -13318,7 +13328,7 @@ function adjustAnnotationsToInsert(annotationsToBeAdjusted, insertStart, insertL
 __name(adjustAnnotationsToInsert, "adjustAnnotationsToInsert");
 function insertSequenceDataAtPositionOrRange(_sequenceDataToInsert, _existingSequenceData, caretPositionOrRange, options = {}) {
   const { maintainOriginSplit } = options;
-  let existingSequenceData = tidyUpSequenceData(_existingSequenceData, options);
+  let existingSequenceData = tidyUpSequenceData(_existingSequenceData, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   const sequenceDataToInsert = tidyUpSequenceData(
     _sequenceDataToInsert,
     options
@@ -13336,6 +13346,7 @@ function insertSequenceDataAtPositionOrRange(_sequenceDataToInsert, _existingSeq
         return acc[type] = [];
       }, {})), {
         sequence: "",
+        doNotRemoveInvalidChars: true,
         proteinSequence: "",
         chromatogramData: void 0
       }),
@@ -21530,7 +21541,7 @@ function getSequenceDataBetweenRange(seqData, range, options = {}) {
   if (!range)
     return seqData;
   const { exclude = {}, excludePartial = {} } = options;
-  const seqDataToUse = tidyUpSequenceData(seqData, options);
+  const seqDataToUse = tidyUpSequenceData(seqData, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   annotationTypes.forEach((type) => {
     delete seqDataToUse[`filtered${lodashExports.startCase(type)}`];
   });
@@ -21581,9 +21592,9 @@ function getSequenceDataBetweenRange(seqData, range, options = {}) {
         }
       });
     });
-    return tidyUpSequenceData(toRet, options);
+    return tidyUpSequenceData(toRet, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   }
-  return tidyUpSequenceData(seqDataToReturn, options);
+  return tidyUpSequenceData(seqDataToReturn, __spreadValues({ doNotRemoveInvalidChars: true }, options));
 }
 __name(getSequenceDataBetweenRange, "getSequenceDataBetweenRange");
 function getAnnotationsBetweenRange(annotationsToBeAdjusted, range, maxLength, shouldExcludePartial) {
@@ -21630,7 +21641,7 @@ function getComplementSequenceAndAnnotations(pSeqObj, options = {}) {
   const newSeqObj = Object.assign({}, seqObj, {
     sequence: getComplementSequenceString(seqObj.sequence, seqObj.isRna)
   });
-  return tidyUpSequenceData(newSeqObj, options);
+  return tidyUpSequenceData(newSeqObj, __spreadValues({ doNotRemoveInvalidChars: true }, options));
 }
 __name(getComplementSequenceAndAnnotations, "getComplementSequenceAndAnnotations");
 function getCutsiteType(restrictionEnzyme) {
@@ -21836,7 +21847,7 @@ __name(getReverseComplementAnnotation, "getReverseComplementAnnotation");
 function getReverseComplementSequenceAndAnnoations(pSeqObj, options = {}) {
   const seqObj = tidyUpSequenceData(
     getSequenceDataBetweenRange(pSeqObj, options.range),
-    options
+    __spreadValues({ doNotRemoveInvalidChars: true }, options)
   );
   const newSeqObj = Object.assign(
     {},
@@ -21856,7 +21867,9 @@ function getReverseComplementSequenceAndAnnoations(pSeqObj, options = {}) {
       return acc;
     }, {})
   );
-  return tidyUpSequenceData(newSeqObj, options);
+  return tidyUpSequenceData(newSeqObj, __spreadValues({
+    doNotRemoveInvalidChars: true
+  }, options));
 }
 __name(getReverseComplementSequenceAndAnnoations, "getReverseComplementSequenceAndAnnoations");
 function getReverseSequenceString(sequence) {

package/index.mjs

@@ -5998,7 +5998,7 @@ lodash.exports;
 var lodashExports = lodash.exports;
 const protein_letters = "ACDEFGHIKLMNPQRSTVWY";
 const protein_letters_withUandX = "ACDEFGHIKLMNPQRSTVWYUX";
-const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO*";
+const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO";
 const ambiguous_dna_letters = "GATCRYWSMKHBVDN";
 const unambiguous_dna_letters = "GATC";
 const ambiguous_rna_letters = "GAUCRYWSMKHBVDN";
@@ -6054,7 +6054,7 @@ const extended_protein_values = {
   Y: "Y",
   Z: "QE",
   "*": "\\*\\.",
-  ".": "\\.\\.",
+  ".": "\\.",
   "-": "\\-"
 };
 const bioData = /* @__PURE__ */ Object.freeze(/* @__PURE__ */ Object.defineProperty({
@@ -12397,9 +12397,12 @@ function getReplaceChars({
   isRna,
   isMixedRnaAndDna
 } = {}) {
-  return isProtein ? {} : isOligo ? {} : isRna ? { t: "u" } : isMixedRnaAndDna ? {} : (
-    //just plain old dna
-    {}
+  return isProtein ? {} : (
+    // {".": "*"}
+    isOligo ? {} : isRna ? { t: "u" } : isMixedRnaAndDna ? {} : (
+      //just plain old dna
+      {}
+    )
   );
 }
 __name(getReplaceChars, "getReplaceChars");
@@ -12670,11 +12673,13 @@ __name(tidyUpSequenceData, "tidyUpSequenceData");
 const getDiffFromSeqs = /* @__PURE__ */ __name((oldData, newData, { ignoreKeys = [] } = {}) => {
   oldData = tidyUpSequenceData(oldData, {
     annotationsAsObjects: true,
-    noTranslationData: true
+    noTranslationData: true,
+    doNotRemoveInvalidChars: true
   });
   newData = tidyUpSequenceData(newData, {
     annotationsAsObjects: true,
-    noTranslationData: true
+    noTranslationData: true,
+    doNotRemoveInvalidChars: true
   });
   [oldData, newData].forEach((d) => {
     [
@@ -12706,7 +12711,10 @@ const patchSeqWithDiff = /* @__PURE__ */ __name((oldData, diff, { ignoreKeys = [
     delete diff[k];
   });
   return jsondiffpatch_umdExports.patch(
-    tidyUpSequenceData(lodashExports.cloneDeep(oldData), { annotationsAsObjects: true }),
+    tidyUpSequenceData(lodashExports.cloneDeep(oldData), {
+      annotationsAsObjects: true,
+      doNotRemoveInvalidChars: true
+    }),
     diff
   );
 }, "patchSeqWithDiff");
@@ -13279,7 +13287,9 @@ function arrayRotate(arr, count) {
 }
 __name(arrayRotate, "arrayRotate");
 function rotateSequenceDataToPosition(sequenceData, caretPosition, options) {
-  const newSequenceData = tidyUpSequenceData(sequenceData, options);
+  const newSequenceData = tidyUpSequenceData(sequenceData, __spreadValues({
+    doNotRemoveInvalidChars: true
+  }, options));
   newSequenceData.sequence = rotateBpsToPosition(
     newSequenceData.sequence,
     caretPosition
@@ -13316,7 +13326,7 @@ function adjustAnnotationsToInsert(annotationsToBeAdjusted, insertStart, insertL
 __name(adjustAnnotationsToInsert, "adjustAnnotationsToInsert");
 function insertSequenceDataAtPositionOrRange(_sequenceDataToInsert, _existingSequenceData, caretPositionOrRange, options = {}) {
   const { maintainOriginSplit } = options;
-  let existingSequenceData = tidyUpSequenceData(_existingSequenceData, options);
+  let existingSequenceData = tidyUpSequenceData(_existingSequenceData, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   const sequenceDataToInsert = tidyUpSequenceData(
     _sequenceDataToInsert,
     options
@@ -13334,6 +13344,7 @@ function insertSequenceDataAtPositionOrRange(_sequenceDataToInsert, _existingSeq
         return acc[type] = [];
       }, {})), {
         sequence: "",
+        doNotRemoveInvalidChars: true,
         proteinSequence: "",
         chromatogramData: void 0
       }),
@@ -21528,7 +21539,7 @@ function getSequenceDataBetweenRange(seqData, range, options = {}) {
   if (!range)
     return seqData;
   const { exclude = {}, excludePartial = {} } = options;
-  const seqDataToUse = tidyUpSequenceData(seqData, options);
+  const seqDataToUse = tidyUpSequenceData(seqData, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   annotationTypes.forEach((type) => {
     delete seqDataToUse[`filtered${lodashExports.startCase(type)}`];
   });
@@ -21579,9 +21590,9 @@ function getSequenceDataBetweenRange(seqData, range, options = {}) {
         }
       });
     });
-    return tidyUpSequenceData(toRet, options);
+    return tidyUpSequenceData(toRet, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   }
-  return tidyUpSequenceData(seqDataToReturn, options);
+  return tidyUpSequenceData(seqDataToReturn, __spreadValues({ doNotRemoveInvalidChars: true }, options));
 }
 __name(getSequenceDataBetweenRange, "getSequenceDataBetweenRange");
 function getAnnotationsBetweenRange(annotationsToBeAdjusted, range, maxLength, shouldExcludePartial) {
@@ -21628,7 +21639,7 @@ function getComplementSequenceAndAnnotations(pSeqObj, options = {}) {
   const newSeqObj = Object.assign({}, seqObj, {
     sequence: getComplementSequenceString(seqObj.sequence, seqObj.isRna)
   });
-  return tidyUpSequenceData(newSeqObj, options);
+  return tidyUpSequenceData(newSeqObj, __spreadValues({ doNotRemoveInvalidChars: true }, options));
 }
 __name(getComplementSequenceAndAnnotations, "getComplementSequenceAndAnnotations");
 function getCutsiteType(restrictionEnzyme) {
@@ -21834,7 +21845,7 @@ __name(getReverseComplementAnnotation, "getReverseComplementAnnotation");
 function getReverseComplementSequenceAndAnnoations(pSeqObj, options = {}) {
   const seqObj = tidyUpSequenceData(
     getSequenceDataBetweenRange(pSeqObj, options.range),
-    options
+    __spreadValues({ doNotRemoveInvalidChars: true }, options)
   );
   const newSeqObj = Object.assign(
     {},
@@ -21854,7 +21865,9 @@ function getReverseComplementSequenceAndAnnoations(pSeqObj, options = {}) {
       return acc;
     }, {})
   );
-  return tidyUpSequenceData(newSeqObj, options);
+  return tidyUpSequenceData(newSeqObj, __spreadValues({
+    doNotRemoveInvalidChars: true
+  }, options));
 }
 __name(getReverseComplementSequenceAndAnnoations, "getReverseComplementSequenceAndAnnoations");
 function getReverseSequenceString(sequence) {

package/index.umd.js

@@ -6002,7 +6002,7 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
   var lodashExports = lodash.exports;
   const protein_letters = "ACDEFGHIKLMNPQRSTVWY";
   const protein_letters_withUandX = "ACDEFGHIKLMNPQRSTVWYUX";
-  const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO*";
+  const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO";
   const ambiguous_dna_letters = "GATCRYWSMKHBVDN";
   const unambiguous_dna_letters = "GATC";
   const ambiguous_rna_letters = "GAUCRYWSMKHBVDN";
@@ -6058,7 +6058,7 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
     Y: "Y",
     Z: "QE",
     "*": "\\*\\.",
-    ".": "\\.\\.",
+    ".": "\\.",
     "-": "\\-"
   };
   const bioData = /* @__PURE__ */ Object.freeze(/* @__PURE__ */ Object.defineProperty({
@@ -12401,9 +12401,12 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
     isRna,
     isMixedRnaAndDna
   } = {}) {
-    return isProtein ? {} : isOligo ? {} : isRna ? { t: "u" } : isMixedRnaAndDna ? {} : (
-      //just plain old dna
-      {}
+    return isProtein ? {} : (
+      // {".": "*"}
+      isOligo ? {} : isRna ? { t: "u" } : isMixedRnaAndDna ? {} : (
+        //just plain old dna
+        {}
+      )
     );
   }
   __name(getReplaceChars, "getReplaceChars");
@@ -12674,11 +12677,13 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
   const getDiffFromSeqs = /* @__PURE__ */ __name((oldData, newData, { ignoreKeys = [] } = {}) => {
     oldData = tidyUpSequenceData(oldData, {
       annotationsAsObjects: true,
-      noTranslationData: true
+      noTranslationData: true,
+      doNotRemoveInvalidChars: true
     });
     newData = tidyUpSequenceData(newData, {
       annotationsAsObjects: true,
-      noTranslationData: true
+      noTranslationData: true,
+      doNotRemoveInvalidChars: true
     });
     [oldData, newData].forEach((d) => {
       [
@@ -12710,7 +12715,10 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
       delete diff[k];
     });
     return jsondiffpatch_umdExports.patch(
-      tidyUpSequenceData(lodashExports.cloneDeep(oldData), { annotationsAsObjects: true }),
+      tidyUpSequenceData(lodashExports.cloneDeep(oldData), {
+        annotationsAsObjects: true,
+        doNotRemoveInvalidChars: true
+      }),
       diff
     );
   }, "patchSeqWithDiff");
@@ -13283,7 +13291,9 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
   }
   __name(arrayRotate, "arrayRotate");
   function rotateSequenceDataToPosition(sequenceData, caretPosition, options) {
-    const newSequenceData = tidyUpSequenceData(sequenceData, options);
+    const newSequenceData = tidyUpSequenceData(sequenceData, __spreadValues({
+      doNotRemoveInvalidChars: true
+    }, options));
     newSequenceData.sequence = rotateBpsToPosition(
       newSequenceData.sequence,
       caretPosition
@@ -13320,7 +13330,7 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
   __name(adjustAnnotationsToInsert, "adjustAnnotationsToInsert");
   function insertSequenceDataAtPositionOrRange(_sequenceDataToInsert, _existingSequenceData, caretPositionOrRange, options = {}) {
     const { maintainOriginSplit } = options;
-    let existingSequenceData = tidyUpSequenceData(_existingSequenceData, options);
+    let existingSequenceData = tidyUpSequenceData(_existingSequenceData, __spreadValues({ doNotRemoveInvalidChars: true }, options));
     const sequenceDataToInsert = tidyUpSequenceData(
       _sequenceDataToInsert,
       options
@@ -13338,6 +13348,7 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
           return acc[type] = [];
         }, {})), {
           sequence: "",
+          doNotRemoveInvalidChars: true,
           proteinSequence: "",
           chromatogramData: void 0
         }),
@@ -21532,7 +21543,7 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
     if (!range)
       return seqData;
     const { exclude = {}, excludePartial = {} } = options;
-    const seqDataToUse = tidyUpSequenceData(seqData, options);
+    const seqDataToUse = tidyUpSequenceData(seqData, __spreadValues({ doNotRemoveInvalidChars: true }, options));
     annotationTypes.forEach((type) => {
       delete seqDataToUse[`filtered${lodashExports.startCase(type)}`];
     });
@@ -21583,9 +21594,9 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
           }
         });
       });
-      return tidyUpSequenceData(toRet, options);
+      return tidyUpSequenceData(toRet, __spreadValues({ doNotRemoveInvalidChars: true }, options));
     }
-    return tidyUpSequenceData(seqDataToReturn, options);
+    return tidyUpSequenceData(seqDataToReturn, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   }
   __name(getSequenceDataBetweenRange, "getSequenceDataBetweenRange");
   function getAnnotationsBetweenRange(annotationsToBeAdjusted, range, maxLength, shouldExcludePartial) {
@@ -21632,7 +21643,7 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
     const newSeqObj = Object.assign({}, seqObj, {
       sequence: getComplementSequenceString(seqObj.sequence, seqObj.isRna)
     });
-    return tidyUpSequenceData(newSeqObj, options);
+    return tidyUpSequenceData(newSeqObj, __spreadValues({ doNotRemoveInvalidChars: true }, options));
   }
   __name(getComplementSequenceAndAnnotations, "getComplementSequenceAndAnnotations");
   function getCutsiteType(restrictionEnzyme) {
@@ -21838,7 +21849,7 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
   function getReverseComplementSequenceAndAnnoations(pSeqObj, options = {}) {
     const seqObj = tidyUpSequenceData(
       getSequenceDataBetweenRange(pSeqObj, options.range),
-      options
+      __spreadValues({ doNotRemoveInvalidChars: true }, options)
     );
     const newSeqObj = Object.assign(
       {},
@@ -21858,7 +21869,9 @@ var __name = (target, value) => __defProp(target, "name", { value, configurable:
         return acc;
       }, {})
     );
-    return tidyUpSequenceData(newSeqObj, options);
+    return tidyUpSequenceData(newSeqObj, __spreadValues({
+      doNotRemoveInvalidChars: true
+    }, options));
   }
   __name(getReverseComplementSequenceAndAnnoations, "getReverseComplementSequenceAndAnnoations");
   function getReverseSequenceString(sequence) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@teselagen/sequence-utils",
-  "version": "0.3.10",
+  "version": "0.3.12",
   "dependencies": {
     "@teselagen/range-utils": "0.3.7",
     "bson-objectid": "^2.0.4",

package/src/bioData.js

@@ -2,7 +2,7 @@
 
 export const protein_letters = "ACDEFGHIKLMNPQRSTVWY";
 export const protein_letters_withUandX = "ACDEFGHIKLMNPQRSTVWYUX";
-export const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO*";
+export const extended_protein_letters = "ACDEFGHIKLMNPQRSTVWYBXZJUO";
 export const ambiguous_dna_letters = "GATCRYWSMKHBVDN";
 export const unambiguous_dna_letters = "GATC";
 export const ambiguous_rna_letters = "GAUCRYWSMKHBVDN";
@@ -60,6 +60,6 @@ export const extended_protein_values = {
   Y: "Y",
   Z: "QE",
   "*": "\\*\\.",
-  ".": "\\.\\.",
+  ".": "\\.",
   "-": "\\-"
 };

package/src/diffUtils.js

@@ -6,11 +6,13 @@ import tidyUpSequenceData from "./tidyUpSequenceData";
 const getDiffFromSeqs = (oldData, newData, { ignoreKeys = [] } = {}) => {
   oldData = tidyUpSequenceData(oldData, {
     annotationsAsObjects: true,
-    noTranslationData: true
+    noTranslationData: true,
+    doNotRemoveInvalidChars: true
   });
   newData = tidyUpSequenceData(newData, {
     annotationsAsObjects: true,
-    noTranslationData: true
+    noTranslationData: true,
+    doNotRemoveInvalidChars: true
   });
 
   [oldData, newData].forEach(d => {
@@ -47,7 +49,10 @@ const patchSeqWithDiff = (oldData, diff, { ignoreKeys = [] } = {}) => {
     delete diff[k];
   });
   return patch(
-    tidyUpSequenceData(cloneDeep(oldData), { annotationsAsObjects: true }),
+    tidyUpSequenceData(cloneDeep(oldData), {
+      annotationsAsObjects: true,
+      doNotRemoveInvalidChars: true
+    }),
     diff
   );
 };

package/src/filterSequenceString.js

@@ -1,7 +1,7 @@
 import {
   ambiguous_dna_letters,
   ambiguous_rna_letters,
-  extended_protein_letters,
+  extended_protein_letters
 } from "./bioData";
 
 export default function filterSequenceString(
@@ -12,14 +12,14 @@ export default function filterSequenceString(
     name,
     isProtein,
     isRna,
-    isMixedRnaAndDna,
+    isMixedRnaAndDna
   } = {}
 ) {
   const acceptedChars = getAcceptedChars({
     isOligo,
     isProtein,
     isRna,
-    isMixedRnaAndDna,
+    isMixedRnaAndDna
   });
   const replaceChars = getReplaceChars({
     isOligo,
@@ -80,7 +80,7 @@ export function getAcceptedChars({
   isOligo,
   isProtein,
   isRna,
-  isMixedRnaAndDna,
+  isMixedRnaAndDna
 } = {}) {
   return isProtein
     ? `${extended_protein_letters.toLowerCase()}}`
@@ -101,8 +101,8 @@ export function getReplaceChars({
 } = {}) {
   return isProtein
     ? {}
-    // {".": "*"}
-    : isOligo
+    : // {".": "*"}
+    isOligo
     ? {}
     : isRna
     ? { t: "u" }

package/src/filterSequenceString.test.js

@@ -50,7 +50,9 @@ describe("filterSequenceString", () => {
       }
     );
     // expect(warnings[0]).toBe(`Replaced "." with "*" 2 times`);
-    expect(warnings[0]).toBe(      'Invalid character(s) detected and removed: 3, 4, 2, ", ", ", ,, ,, ., ., / ');
+    expect(warnings[0]).toBe(
+      'Invalid character(s) detected and removed: 3, 4, 2, ", ", ", ,, ,, ., ., / '
+    );
     expect(str).toBe("bbbxtgalmfwkqespvicyhrnd");
   });
   it("when isProtein: true, should handle upper case letters", () => {
@@ -60,11 +62,27 @@ describe("filterSequenceString", () => {
     expect(warnings.length).toBe(0);
     expect(str).toBe("xtgalmfWKQEspvicyhrnd");
   });
+
+  it("when isProtein: true it should not filter this aa seq", () => {
+    const [str] = filterSequenceString(
+      "mhhhhhhgsgsmledlkrqvleanlalpkhnlasgssghvsavdrergvfviapsgvdfrimtaddmvvvsietgevvegekppaedtpthrllyqafpsiggivhthsrhatiwaqagqsipatgtthadhfygtipctrkmtdaeingeyewetgnvivetfekqgidaaqmpgvlvhshgpfawgknaedavhnaivleevaymgifcrqlapqlpdmqqtllnkhylrkhgakayygq",
+      {
+        isProtein: true
+      }
+    );
+
+    expect(str).toBe(
+      `mhhhhhhgsgsmledlkrqvleanlalpkhnlasgssghvsavdrergvfviapsgvdfrimtaddmvvvsietgevvegekppaedtpthrllyqafpsiggivhthsrhatiwaqagqsipatgtthadhfygtipctrkmtdaeingeyewetgnvivetfekqgidaaqmpgvlvhshgpfawgknaedavhnaivleevaymgifcrqlapqlpdmqqtllnkhylrkhgakayygq`
+    );
+  });
   it("when isProtein: true, it should convert . to *", () => {
-    const [str] = filterSequenceString('BXZJUO*bbb342"""xtgalbmfwkqespvicyhrnd,,../', {
-      isProtein: true,
-    });
+    const [str] = filterSequenceString(
+      'BXZJUO*bbb342"""xtgalbmfwkqespvicyhrnd,,../',
+      {
+        isProtein: true
+      }
+    );
 
-    expect(str).toBe("BXZJUO*bbbxtgalbmfwkqespvicyhrnd");
+    expect(str).toBe("BXZJUObbbxtgalbmfwkqespvicyhrnd");
   });
 });

package/src/getComplementSequenceAndAnnotations.js

@@ -15,5 +15,5 @@ export default function getComplementSequenceAndAnnotations(
   const newSeqObj = Object.assign({}, seqObj, {
     sequence: getComplementSequenceString(seqObj.sequence, seqObj.isRna)
   });
-  return tidyUpSequenceData(newSeqObj, options);
+  return tidyUpSequenceData(newSeqObj, {doNotRemoveInvalidChars: true,...options});
 }

package/src/getReverseComplementSequenceAndAnnotations.js

@@ -13,7 +13,7 @@ export default function getReverseComplementSequenceAndAnnoations(
 ) {
   const seqObj = tidyUpSequenceData(
     getSequenceDataBetweenRange(pSeqObj, options.range),
-    options
+    { doNotRemoveInvalidChars: true, ...options }
   );
   const newSeqObj = Object.assign(
     {},
@@ -33,5 +33,8 @@ export default function getReverseComplementSequenceAndAnnoations(
       return acc;
     }, {})
   );
-  return tidyUpSequenceData(newSeqObj, options);
+  return tidyUpSequenceData(newSeqObj, {
+    doNotRemoveInvalidChars: true,
+    ...options
+  });
 }

package/src/getSequenceDataBetweenRange.js

@@ -16,7 +16,7 @@ export default function getSequenceDataBetweenRange(
 ) {
   if (!range) return seqData;
   const { exclude = {}, excludePartial = {} } = options;
-  const seqDataToUse = tidyUpSequenceData(seqData, options);
+  const seqDataToUse = tidyUpSequenceData(seqData, {doNotRemoveInvalidChars: true,...options});
   annotationTypes.forEach(type => {
     delete seqDataToUse[`filtered${startCase(type)}`];
   });
@@ -78,10 +78,10 @@ export default function getSequenceDataBetweenRange(
         }
       });
     });
-    return tidyUpSequenceData(toRet, options);
+    return tidyUpSequenceData(toRet, {doNotRemoveInvalidChars: true,...options});
   }
 
-  return tidyUpSequenceData(seqDataToReturn, options);
+  return tidyUpSequenceData(seqDataToReturn, {doNotRemoveInvalidChars: true,...options});
 }
 
 function getAnnotationsBetweenRange(

package/src/insertSequenceDataAtPositionOrRange.js

@@ -18,7 +18,7 @@ export default function insertSequenceDataAtPositionOrRange(
   //when inserting new seq, n bps of the new seq should go in before the origin and the rest should be
   //inserted at the sequence start
   const { maintainOriginSplit } = options;
-  let existingSequenceData = tidyUpSequenceData(_existingSequenceData, options);
+  let existingSequenceData = tidyUpSequenceData(_existingSequenceData, {doNotRemoveInvalidChars: true,...options});
   const sequenceDataToInsert = tidyUpSequenceData(
     _sequenceDataToInsert,
     options
@@ -48,6 +48,7 @@ export default function insertSequenceDataAtPositionOrRange(
           return (acc[type] = []);
         }, {}),
         sequence: "",
+        doNotRemoveInvalidChars: true,
         proteinSequence: "",
         chromatogramData: undefined
       },

package/src/insertSequenceDataAtPositionOrRange.test.js

@@ -135,7 +135,7 @@ describe("insertSequenceData", () => {
   });
   it("inserts characters at correct origin spanning range with {maintainOriginSplit: true} option", () => {
     const sequenceToInsert = {
-      sequence: "xrrrrry",
+      sequence: "crrrrry",
       //         fffffff
       features: [{ name: "feat1", start: 0, end: 6 }]
     };
@@ -154,7 +154,7 @@ describe("insertSequenceData", () => {
         maintainOriginSplit: true
       }
     );
-    postInsertSeq.sequence.should.equal("rrrryagagaxr");
+    postInsertSeq.sequence.should.equal("rrrryagagacr");
     //                                   fffff fff ff
     postInsertSeq.features.should.containSubset([
       { name: "feat1", start: 10, end: 4 },

package/src/rotateSequenceDataToPosition.js

@@ -9,7 +9,10 @@ export default function rotateSequenceDataToPosition(
   caretPosition,
   options
 ) {
-  const newSequenceData = tidyUpSequenceData(sequenceData, options);
+  const newSequenceData = tidyUpSequenceData(sequenceData, {
+    doNotRemoveInvalidChars: true,
+    ...options
+  });
 
   //update the sequence
   newSequenceData.sequence = rotateBpsToPosition(

package/src/tidyUpSequenceData.test.js

@@ -6,28 +6,25 @@ chai.use(chaiSubset);
 chai.should();
 describe("tidyUpSequenceData", () => {
   it("should remove invalid chars by default, while handling annotation start,end (and location start,end) truncation correctly", () => {
-    const res = tidyUpSequenceData(
-      {
-        sequence: "http://localhost:3344/Standalone",
-        features: [
-          {
-            start: 3,
-            end: 20,
-            locations: [
-              {
-                start: "3", //this should be converted to an int :)
-                end: 5
-              },
-              {
-                start: 10,
-                end: 20
-              }
-            ]
-          }
-        ]
-      },
-      
-    );
+    const res = tidyUpSequenceData({
+      sequence: "http://localhost:3344/Standalone",
+      features: [
+        {
+          start: 3,
+          end: 20,
+          locations: [
+            {
+              start: "3", //this should be converted to an int :)
+              end: 5
+            },
+            {
+              start: 10,
+              end: 20
+            }
+          ]
+        }
+      ]
+    });
     res.should.containSubset({
       sequence: "httcahstStandan",
       circular: false,
@@ -64,22 +61,22 @@ describe("tidyUpSequenceData", () => {
       },
       { convertAnnotationsFromAAIndices: true }
     );
-    
+
     res.should.containSubset({
       aminoAcidDataForEachBaseOfDNA: [],
       isProtein: true,
-      size: 57, //size should refer to the DNA length
-      proteinSize: 19, //proteinSize should refer to the amino acid length
-      sequence: "ggngcnggnathtgacaytggggngcnggngcnytngcnwsnhtnggnytnhtntrr", //degenerate sequence
-      proteinSequence: "gagiuhwgagalasjglj*",
+      size: 54, //size should refer to the DNA length
+      proteinSize: 18, //proteinSize should refer to the amino acid length
+      sequence: "ggngcnggnathtgacaytggggngcnggngcnytngcnwsnhtnggnytnhtn", //degenerate sequence
+      proteinSequence: "gagiuhwgagalasjglj",
       circular: false,
       features: [
         { start: 9, end: 32, forward: true },
-        { start: 30, end: 56, forward: true },
+        { start: 30, end: 53, forward: true },
         {
           name: "iDon'tFit",
-          start: 54,
-          end: 56,
+          start: 51,
+          end: 53,
           forward: true
         }
       ]