@teselagen/sequence-utils 0.3.41 → 0.3.42-beta.1

package/tidyUpAnnotation.d.ts

@@ -1,11 +1,13 @@
-export default function tidyUpAnnotation(_annotation: any, { sequenceData, convertAnnotationsFromAAIndices, annotationType, provideNewIdsForAnnotations, doNotProvideIdsForAnnotations, messages, mutative, allowNonStandardGenbankTypes, featureTypes }: {
-    sequenceData?: {} | undefined;
-    convertAnnotationsFromAAIndices: any;
-    annotationType: any;
-    provideNewIdsForAnnotations: any;
-    doNotProvideIdsForAnnotations: any;
-    messages?: never[] | undefined;
-    mutative: any;
-    allowNonStandardGenbankTypes: any;
-    featureTypes: any;
-}): any;
+import { Annotation, SequenceData } from './types';
+export interface TidyUpAnnotationOptions {
+    sequenceData?: Partial<SequenceData>;
+    convertAnnotationsFromAAIndices?: boolean;
+    annotationType?: string;
+    provideNewIdsForAnnotations?: boolean;
+    doNotProvideIdsForAnnotations?: boolean;
+    messages?: string[];
+    mutative?: boolean;
+    allowNonStandardGenbankTypes?: boolean;
+    featureTypes?: string[];
+}
+export default function tidyUpAnnotation(_annotation: Annotation, { sequenceData, convertAnnotationsFromAAIndices, annotationType, provideNewIdsForAnnotations, doNotProvideIdsForAnnotations, messages, mutative, allowNonStandardGenbankTypes, featureTypes }: TidyUpAnnotationOptions): false | Annotation;

package/tidyUpSequenceData.d.ts

@@ -1 +1,18 @@
-export default function tidyUpSequenceData(pSeqData: any, options?: {}): any;
+import { FilterSequenceStringOptions } from './filterSequenceString';
+import { SequenceData } from './types';
+export interface TidyUpSequenceDataOptions {
+    annotationsAsObjects?: boolean;
+    logMessages?: boolean;
+    doNotRemoveInvalidChars?: boolean;
+    additionalValidChars?: string;
+    noTranslationData?: boolean;
+    getAcceptedInsertChars?: (options: FilterSequenceStringOptions) => string;
+    includeProteinSequence?: boolean;
+    doNotProvideIdsForAnnotations?: boolean;
+    noCdsTranslations?: boolean;
+    convertAnnotationsFromAAIndices?: boolean;
+    topLevelSeqData?: Partial<SequenceData>;
+    allowNonStandardGenbankTypes?: boolean;
+    isMixedRnaAndDna?: boolean;
+}
+export default function tidyUpSequenceData(pSeqData: Partial<SequenceData>, options?: TidyUpSequenceDataOptions): SequenceData;

package/types.d.ts

@@ -0,0 +1,96 @@
+import { Range } from '../../range-utils/src/index.ts';
+export interface Annotation extends Range {
+    id?: string | number;
+    name?: string;
+    forward?: boolean;
+    strand?: number | string;
+    type?: string;
+    color?: string;
+    notes?: Record<string, unknown> | string;
+    annotationTypePlural?: string;
+    translationType?: string;
+    [key: string]: unknown;
+}
+export interface SequenceData {
+    id?: string;
+    sequence: string;
+    proteinSequence?: string;
+    circular?: boolean;
+    isProtein?: boolean;
+    isRna?: boolean;
+    size?: number;
+    proteinSize?: number;
+    name?: string;
+    description?: string;
+    features?: Annotation[];
+    parts?: Annotation[];
+    translations?: Annotation[];
+    primers?: Annotation[];
+    cutsites?: Annotation[];
+    orfs?: Annotation[];
+    guides?: Annotation[];
+    noSequence?: boolean;
+    sequenceTypeCode?: string;
+    aminoAcidDataForEachBaseOfDNA?: unknown[];
+    chromatogramData?: ChromatogramData;
+    [key: string]: unknown;
+}
+export interface ChromatogramData {
+    baseTraces?: unknown[];
+    baseCalls?: unknown[];
+    qualNums?: unknown[];
+    basePos?: unknown[];
+    [key: string]: unknown;
+}
+export interface RestrictionEnzyme {
+    name: string;
+    site: string;
+    forwardRegex: string;
+    reverseRegex: string;
+    topSnipOffset?: number;
+    bottomSnipOffset?: number;
+    cutType?: number;
+    usForward?: number;
+    usReverse?: number;
+    [key: string]: unknown;
+}
+export interface CutSite extends Annotation {
+    topSnipPosition: number | null;
+    bottomSnipPosition: number | null;
+    overhangSize: number;
+    overhangBps?: string;
+    restrictionEnzyme: RestrictionEnzyme;
+    upstreamTopSnip?: number | null;
+    upstreamBottomSnip?: number | null;
+    topSnipBeforeBottom?: boolean;
+    upstreamTopBeforeBottom?: boolean;
+    cutType?: number;
+    cutsTwice?: boolean;
+    recognitionSiteRange?: Range;
+    isOverhangIncludedInFragmentSize?: boolean;
+    [key: string]: unknown;
+}
+export interface DigestFragment extends Range {
+    isFormedFromLinearEnd?: boolean;
+    madeFromOneCutsite?: boolean;
+    size: number;
+    cut1: CutSite;
+    cut2: CutSite;
+    id: string;
+    name: string;
+    onFragmentSelect?: () => void;
+    [key: string]: unknown;
+}
+export interface AminoAcidData {
+    fullCodon: boolean | null;
+    aminoAcid: {
+        value: string;
+    } | null;
+    aminoAcidIndex: number | null;
+    positionInCodon?: number | null;
+    sequenceIndex?: number | null;
+    codonRange?: {
+        start: number;
+        end: number;
+    } | null;
+}

package/addGapsToSeqReads.test.d.ts

@@ -1 +0,0 @@
-export {};

package/adjustBpsToReplaceOrInsert.test.d.ts

@@ -1 +0,0 @@
-export {};

package/aminoAcidToDnaRna.test.d.ts

@@ -1 +0,0 @@
-export {};

package/annotateSingleSeq.test.d.ts

@@ -1 +0,0 @@
-export {};

package/autoAnnotate.test.d.ts

@@ -1 +0,0 @@
-export {};

package/calculateEndStability.test.d.ts

@@ -1 +0,0 @@
-export {};

package/calculateNebTa.test.d.ts

@@ -1 +0,0 @@
-export {};

package/calculateNebTm.test.d.ts

@@ -1 +0,0 @@
-export {};

package/calculatePercentGC.test.d.ts

@@ -1 +0,0 @@
-export {};

package/calculateSantaLuciaTm.test.d.ts

@@ -1 +0,0 @@
-export {};

package/calculateTm.test.d.ts

@@ -1 +0,0 @@
-export {};

package/computeDigestFragments.test.d.ts

@@ -1 +0,0 @@
-export {};

package/condensePairwiseAlignmentDifferences.test.d.ts

@@ -1 +0,0 @@
-export {};

package/convertAACaretPositionOrRangeToDna.test.d.ts

@@ -1 +0,0 @@
-export {};

package/convertDnaCaretPositionOrRangeToAA.test.d.ts

@@ -1 +0,0 @@
-export {};

package/cutSequenceByRestrictionEnzyme.test.d.ts

@@ -1 +0,0 @@
-export {};

package/deleteSequenceDataAtRange.test.d.ts

@@ -1 +0,0 @@
-export {};

package/diffUtils.test.d.ts

@@ -1 +0,0 @@
-export {};

package/doesEnzymeChopOutsideOfRecognitionSite.test.d.ts

@@ -1 +0,0 @@
-export {};

package/featureTypesAndColors.test.d.ts

@@ -1 +0,0 @@
-export {};

package/filterSequenceString.test.d.ts

@@ -1 +0,0 @@
-export {};

package/findApproxMatches.test.d.ts

@@ -1 +0,0 @@
-export {};

package/findNearestRangeOfSequenceOverlapToPosition.test.d.ts

@@ -1 +0,0 @@
-export {};

package/findSequenceMatches.test.d.ts

@@ -1 +0,0 @@
-export {};

package/generateSequenceData.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getAllInsertionsInSeqReads.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getAminoAcidDataForEachBaseOfDna.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getAminoAcidStringFromSequenceString.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getComplementSequenceString.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getDigestFragmentsForRestrictionEnzymes.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getInsertBetweenVals.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getLeftAndRightOfSequenceInRangeGivenPosition.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getMassofAaString.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getOrfsFromSequence.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getOverlapBetweenTwoSequences.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getPossiblePartsFromSequenceAndEnzymes.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getReverseAminoAcidStringFromSequenceString.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getReverseComplementAnnotation.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getReverseComplementSequenceAndAnnotations.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getReverseComplementSequenceString.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getReverseSequenceString.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getSequenceDataBetweenRange.test.d.ts

@@ -1 +0,0 @@
-export {};

package/getVirtualDigest.test.d.ts

@@ -1 +0,0 @@
-export {};

package/guessIfSequenceIsDnaAndNotProtein.test.d.ts

@@ -1 +0,0 @@
-export {};

package/index.test.d.ts

@@ -1 +0,0 @@
-export {};

package/insertGapsIntoRefSeq.test.d.ts

@@ -1 +0,0 @@
-export {};

package/insertSequenceDataAtPosition.test.d.ts

@@ -1 +0,0 @@
-export {};

package/insertSequenceDataAtPositionOrRange.test.d.ts

@@ -1 +0,0 @@
-export {};

package/mapAnnotationsToRows.test.d.ts

@@ -1 +0,0 @@
-export {};

package/prepareCircularViewData.test.d.ts

@@ -1 +0,0 @@
-export {};

package/prepareRowData.test.d.ts

@@ -1 +0,0 @@
-export {};

package/rotateBpsToPosition.test.d.ts

@@ -1 +0,0 @@
-export {};

package/rotateSequenceDataToPosition.test.d.ts

@@ -1 +0,0 @@
-export {};

package/src/cutSequenceByRestrictionEnzyme.js

@@ -1,301 +0,0 @@
-import { assign } from "lodash-es";
-import shortid from "shortid";
-import getReverseComplementSequenceString from "./getReverseComplementSequenceString";
-
-import {
-  getSequenceWithinRange,
-  normalizePositionByRangeLength,
-  reversePositionInRange
-} from "@teselagen/range-utils";
-
-export default function cutSequenceByRestrictionEnzyme(
-  pSequence,
-  circular,
-  restrictionEnzyme
-) {
-  if (
-    restrictionEnzyme.forwardRegex.length === 0 ||
-    restrictionEnzyme.reverseRegex.length === 0
-  ) {
-    const returnArray = [];
-    returnArray.error =
-      "Cannot cut sequence. Enzyme restriction site must be at least 1 bp long.";
-    return returnArray;
-  }
-  const forwardRegExpPattern = new RegExp(restrictionEnzyme.forwardRegex, "ig");
-  const sequence = pSequence;
-
-  const cutsitesForward = cutSequence(
-    forwardRegExpPattern,
-    restrictionEnzyme,
-    sequence,
-    circular
-  );
-  let cutsitesReverse = [];
-  if (restrictionEnzyme.forwardRegex !== restrictionEnzyme.reverseRegex) {
-    const revSequence = getReverseComplementSequenceString(sequence);
-    cutsitesReverse = cutSequence(
-      forwardRegExpPattern,
-      restrictionEnzyme,
-      revSequence,
-      circular
-    );
-    cutsitesReverse = cutsitesReverse.map(cutsite => {
-      return reverseAllPositionsOfCutsite(cutsite, sequence.length);
-    });
-  }
-  return cutsitesForward.concat(cutsitesReverse);
-
-  function reverseAllPositionsOfCutsite(cutsite, rangeLength) {
-    cutsite.start = reversePositionInRange(cutsite.start, rangeLength, false);
-    cutsite.end = reversePositionInRange(cutsite.end, rangeLength, false);
-    cutsite.topSnipPosition = reversePositionInRange(
-      cutsite.topSnipPosition,
-      rangeLength,
-      true
-    );
-    cutsite.bottomSnipPosition = reversePositionInRange(
-      cutsite.bottomSnipPosition,
-      rangeLength,
-      true
-    );
-    if (cutsite.cutsTwice) {
-      cutsite.upstreamTopSnip = reversePositionInRange(
-        cutsite.upstreamTopSnip,
-        rangeLength,
-        true
-      );
-      cutsite.upstreamBottomSnip = reversePositionInRange(
-        cutsite.upstreamBottomSnip,
-        rangeLength,
-        true
-      );
-    }
-    cutsite.recognitionSiteRange.start = reversePositionInRange(
-      cutsite.recognitionSiteRange.start,
-      rangeLength,
-      false
-    );
-    cutsite.recognitionSiteRange.end = reversePositionInRange(
-      cutsite.recognitionSiteRange.end,
-      rangeLength,
-      false
-    );
-    return assign({}, cutsite, {
-      start: cutsite.end,
-      end: cutsite.start,
-      overhangBps: getReverseComplementSequenceString(cutsite.overhangBps),
-      topSnipPosition: cutsite.bottomSnipPosition,
-      bottomSnipPosition: cutsite.topSnipPosition,
-      upstreamTopSnip: cutsite.upstreamBottomSnip,
-      upstreamBottomSnip: cutsite.upstreamTopSnip,
-      upstreamTopBeforeBottom: !!cutsite.upstreamTopBeforeBottom,
-      topSnipBeforeBottom: !!cutsite.topSnipBeforeBottom,
-      recognitionSiteRange: {
-        start: cutsite.recognitionSiteRange.end,
-        end: cutsite.recognitionSiteRange.start
-      },
-      forward: false
-    });
-  }
-}
-
-function cutSequence(
-  forwardRegExpPattern,
-  restrictionEnzyme,
-  sequence,
-  circular
-) {
-  const restrictionCutSites = [];
-  let restrictionCutSite;
-  const recognitionSiteLength = restrictionEnzyme.site.length;
-  const originalSequence = sequence;
-  const originalSequenceLength = sequence.length;
-  if (circular) {
-    //if the sequence is circular, we send in double the sequence
-    //we'll deduplicate the results afterwards!
-    sequence += sequence;
-  }
-  const currentSequenceLength = sequence.length;
-
-  let matchIndex = sequence.search(forwardRegExpPattern);
-  let startIndex = 0;
-  let subSequence = sequence;
-
-  while (matchIndex !== -1) {
-    const recognitionSiteRange = {};
-    let start; //start and end should fully enclose the enzyme snips and the recognition site!
-    let end;
-    let upstreamTopSnip = null; //upstream top snip position
-    let upstreamBottomSnip = null; //upstream bottom snip position
-    let upstreamTopBeforeBottom = false;
-    let topSnipPosition = null; //downstream top snip position
-    let bottomSnipPosition = null; //downstream bottom snip position
-    let topSnipBeforeBottom = false;
-
-    let fitsWithinSequence = false;
-    // if (matchIndex + startIndex + recognitionSiteLength - 1 >= sequence.length) { // subSequence is too short
-    //     break;
-    // }
-
-    recognitionSiteRange.start = matchIndex + startIndex;
-    start = recognitionSiteRange.start; //this might change later on!
-
-    recognitionSiteRange.end =
-      matchIndex + recognitionSiteLength - 1 + startIndex;
-    end = recognitionSiteRange.end; //this might change later on!
-
-    //we need to get the snip sites, top and bottom for each of these cut sites
-    //as well as all of the bp's between the snip sites
-
-    //if the cutsite is type 1, it cuts both upstream and downstream of its recognition site (cutsite type 0's cut only downstream)
-    if (restrictionEnzyme.cutType === 1) {
-      //double cutter, add upstream cutsite here
-      upstreamTopSnip = recognitionSiteRange.end - restrictionEnzyme.usForward;
-      upstreamBottomSnip =
-        recognitionSiteRange.end - restrictionEnzyme.usReverse;
-      if (upstreamTopSnip >= 0 && upstreamBottomSnip >= 0) {
-        fitsWithinSequence = true;
-        if (upstreamTopSnip < upstreamBottomSnip) {
-          if (start > upstreamTopSnip) {
-            start = upstreamTopSnip + 1;
-          }
-          upstreamTopBeforeBottom = true;
-        } else {
-          if (start > upstreamBottomSnip) {
-            start = upstreamBottomSnip + 1;
-          }
-        }
-        upstreamTopSnip = normalizePositionByRangeLength(
-          upstreamTopSnip,
-          originalSequenceLength,
-          true
-        );
-        upstreamBottomSnip = normalizePositionByRangeLength(
-          upstreamBottomSnip,
-          originalSequenceLength,
-          true
-        );
-      } else {
-        upstreamTopSnip = null;
-        upstreamBottomSnip = null;
-      }
-    }
-
-    //add downstream cutsite here
-    topSnipPosition =
-      recognitionSiteRange.start + restrictionEnzyme.topSnipOffset;
-    bottomSnipPosition =
-      recognitionSiteRange.start + restrictionEnzyme.bottomSnipOffset;
-    if (
-      bottomSnipPosition <= currentSequenceLength &&
-      topSnipPosition <= currentSequenceLength
-    ) {
-      fitsWithinSequence = true;
-      if (topSnipPosition > bottomSnipPosition) {
-        if (topSnipPosition > recognitionSiteRange.end) {
-          end = topSnipPosition - 1;
-        }
-      } else {
-        if (bottomSnipPosition > recognitionSiteRange.end) {
-          end = bottomSnipPosition - 1;
-        }
-        topSnipBeforeBottom = true;
-      }
-      topSnipPosition = normalizePositionByRangeLength(
-        topSnipPosition,
-        originalSequenceLength,
-        true
-      );
-      bottomSnipPosition = normalizePositionByRangeLength(
-        bottomSnipPosition,
-        originalSequenceLength,
-        true
-      );
-    } else {
-      topSnipPosition = null;
-      bottomSnipPosition = null;
-    }
-
-    if (
-      fitsWithinSequence &&
-      start >= 0 &&
-      end >= 0 &&
-      start < originalSequenceLength &&
-      end < currentSequenceLength
-    ) {
-      //only push cutsites onto the array if they are fully contained within the boundaries of the sequence!
-      //and they aren't duplicated
-      start = normalizePositionByRangeLength(
-        start,
-        originalSequenceLength,
-        false
-      );
-      end = normalizePositionByRangeLength(end, originalSequenceLength, false);
-      recognitionSiteRange.start = normalizePositionByRangeLength(
-        recognitionSiteRange.start,
-        originalSequenceLength,
-        false
-      );
-      recognitionSiteRange.end = normalizePositionByRangeLength(
-        recognitionSiteRange.end,
-        originalSequenceLength,
-        false
-      );
-      let cutRange = {
-        start: -1,
-        end: -1
-      };
-
-      if (topSnipPosition !== bottomSnipPosition) {
-        //there is only a cut range if the snips don't snip at the exact same spot on top and bottom
-        cutRange = topSnipBeforeBottom
-          ? {
-              start: topSnipPosition,
-              end: normalizePositionByRangeLength(
-                bottomSnipPosition - 1,
-                originalSequenceLength
-              )
-            }
-          : {
-              start: bottomSnipPosition,
-              end: normalizePositionByRangeLength(
-                topSnipPosition - 1,
-                originalSequenceLength
-              )
-            };
-      }
-      const overhangBps = getSequenceWithinRange(cutRange, originalSequence);
-
-      restrictionCutSite = {
-        id: shortid(),
-        start,
-        end,
-        topSnipPosition,
-        bottomSnipPosition,
-        topSnipBeforeBottom,
-        overhangBps,
-        overhangSize: overhangBps.length,
-        upstreamTopBeforeBottom,
-        upstreamTopSnip,
-        annotationTypePlural: "cutsites",
-        upstreamBottomSnip,
-        recognitionSiteRange,
-        forward: true,
-        name: restrictionEnzyme.name,
-        restrictionEnzyme
-      };
-      restrictionCutSites.push(restrictionCutSite);
-    }
-
-    // Make sure that we always store the previous match index to ensure
-    // that we are always storing indices relative to the whole sequence,
-    // not just the subSequence.
-    startIndex = startIndex + matchIndex + 1;
-
-    // Search again on subSequence, starting from the index of the last match + 1.
-    subSequence = sequence.substring(startIndex, sequence.length);
-    matchIndex = subSequence.search(forwardRegExpPattern);
-  }
-  return restrictionCutSites;
-}

package/src/deleteSequenceDataAtRange.js

@@ -1,5 +0,0 @@
-import insertSequenceDataAtPositionOrRange from "./insertSequenceDataAtPositionOrRange";
-
-export default function deleteSequenceDataAtRange(sequenceData, range) {
-  return insertSequenceDataAtPositionOrRange({}, sequenceData, range);
-}