@teselagen/sequence-utils 0.3.7 → 0.3.9

package/src/proteinAlphabet.js

@@ -6,7 +6,7 @@ const proteinAlphabet = {
     hydrophobicity: 1.8,
     colorByFamily: "#00FFFF",
     color: "hsl(327.3, 100%, 69%)",
-    mass: 89.1
+    mass: 71.0779
   },
   R: {
     value: "R",
@@ -15,8 +15,9 @@ const proteinAlphabet = {
     hydrophobicity: -4.5,
     colorByFamily: "#FFC0CB",
     color: "hsl(258.1, 100%, 69%)",
-    mass: 174.2
+    mass: 156.18568
   },
+
   N: {
     value: "N",
     name: "Asparagine",
@@ -24,7 +25,7 @@ const proteinAlphabet = {
     hydrophobicity: -3.5,
     colorByFamily: "#D3D3D3",
     color: "hsl(268.9, 100%, 69%)",
-    mass: 132.1
+    mass: 114.10264
   },
   D: {
     value: "D",
@@ -33,7 +34,7 @@ const proteinAlphabet = {
     hydrophobicity: -3.5,
     colorByFamily: "#EE82EE",
     color: "hsl(268.9, 100%, 69%)",
-    mass: 133.1
+    mass: 115.0874
   },
   C: {
     value: "C",
@@ -42,8 +43,9 @@ const proteinAlphabet = {
     hydrophobicity: 2.5,
     colorByFamily: "#FFFF00",
     color: "hsl(335.1, 100%, 69%)",
-    mass: 121.2
+    mass: 103.1429
   },
+
   E: {
     value: "E",
     name: "Glutamic acid",
@@ -51,7 +53,7 @@ const proteinAlphabet = {
     hydrophobicity: -3.5,
     colorByFamily: "#EE82EE",
     color: "hsl(268.9, 100%, 69%)",
-    mass: 147.1
+    mass: 129.11398
   },
   Q: {
     value: "Q",
@@ -60,7 +62,7 @@ const proteinAlphabet = {
     hydrophobicity: -3.5,
     colorByFamily: "#D3D3D3",
     color: "hsl(268.9, 100%, 69%)",
-    mass: 146.2
+    mass: 128.12922
   },
   G: {
     value: "G",
@@ -69,8 +71,9 @@ const proteinAlphabet = {
     hydrophobicity: -0.4,
     colorByFamily: "#00FFFF",
     color: "hsl(303.1, 100%, 69%)",
-    mass: 75.1
+    mass: 57.05132
   },
+
   H: {
     value: "H",
     name: "Histidine",
@@ -78,8 +81,9 @@ const proteinAlphabet = {
     hydrophobicity: -3.2,
     colorByFamily: "#FFC0CB",
     color: "hsl(272.2, 100%, 69%)",
-    mass: 155.2
+    mass: 137.13928
   },
+
   I: {
     value: "I",
     name: "Isoleucine ",
@@ -87,7 +91,7 @@ const proteinAlphabet = {
     hydrophobicity: 4.5,
     colorByFamily: "#00FFFF",
     color: "hsl(356.9, 100%, 69%)",
-    mass: 131.2
+    mass: 113.15764
   },
   L: {
     value: "L",
@@ -96,7 +100,7 @@ const proteinAlphabet = {
     hydrophobicity: 3.8,
     colorByFamily: "#00FFFF",
     color: "hsl(349.4, 100%, 69%)",
-    mass: 131.2
+    mass: 113.15764
   },
   K: {
     value: "K",
@@ -105,8 +109,9 @@ const proteinAlphabet = {
     hydrophobicity: -3.9,
     colorByFamily: "#FFC0CB",
     color: "hsl(264.7, 100%, 69%)",
-    mass: 146.2
+    mass: 128.17228
   },
+
   M: {
     value: "M",
     name: "Methionine",
@@ -114,7 +119,7 @@ const proteinAlphabet = {
     hydrophobicity: 1.9,
     colorByFamily: "#FFFF00",
     color: "hsl(328.5, 100%, 69%)",
-    mass: 149.2
+    mass: 131.19606
   },
   F: {
     value: "F",
@@ -123,7 +128,7 @@ const proteinAlphabet = {
     hydrophobicity: 2.8,
     colorByFamily: "#FFA500",
     color: "hsl(338.4, 100%, 69%)",
-    mass: 165.2
+    mass: 147.17386
   },
   P: {
     value: "P",
@@ -132,7 +137,7 @@ const proteinAlphabet = {
     hydrophobicity: -1.6,
     colorByFamily: "#00FFFF",
     color: "hsl(289.9, 100%, 69%)",
-    mass: 115.1
+    mass: 97.11518
   },
   S: {
     value: "S",
@@ -141,7 +146,7 @@ const proteinAlphabet = {
     hydrophobicity: -0.8,
     colorByFamily: "#90EE90",
     color: "hsl(298.6, 100%, 69%)",
-    mass: 105.1
+    mass: 87.0773
   },
   T: {
     value: "T",
@@ -150,7 +155,7 @@ const proteinAlphabet = {
     hydrophobicity: -0.7,
     colorByFamily: "#90EE90",
     color: "hsl(299.8, 100%, 69%)",
-    mass: 119.1
+    mass: 101.10388
   },
   U: {
     value: "U",
@@ -158,7 +163,7 @@ const proteinAlphabet = {
     threeLettersName: "Sec",
     colorByFamily: "#FF0000",
     color: "hsl(0, 100%, 69%)",
-    mass: 168.1
+    mass: 150.3079
   },
   W: {
     value: "W",
@@ -167,7 +172,7 @@ const proteinAlphabet = {
     hydrophobicity: -0.9,
     colorByFamily: "#FFA500",
     color: "hsl(297.6, 100%, 69%)",
-    mass: 204.2
+    mass: 186.2099
   },
   Y: {
     value: "Y",
@@ -176,7 +181,7 @@ const proteinAlphabet = {
     hydrophobicity: -1.3,
     colorByFamily: "#FFA500",
     color: "hsl(293.2, 100%, 69%)",
-    mass: 181.2
+    mass: 163.17326
   },
   V: {
     value: "V",
@@ -185,7 +190,7 @@ const proteinAlphabet = {
     hydrophobicity: 4.2,
     colorByFamily: "#00FFFF",
     color: "hsl(353.6, 100%, 69%)",
-    mass: 117.1
+    mass: 99.13106
   },
   "*": {
     value: "*",
@@ -210,7 +215,7 @@ const proteinAlphabet = {
     threeLettersName: "Gap",
     colorByFamily: "#FF0000",
     color: "hsl(0, 100%, 69%)",
-    mass: 0,
+    mass: 0
   },
   B: {
     value: "B",
@@ -220,7 +225,7 @@ const proteinAlphabet = {
     isAmbiguous: true,
     name: "B",
     aliases: "ND",
-    mass: 0,
+    mass: 0
   },
   J: {
     value: "J",
@@ -230,7 +235,7 @@ const proteinAlphabet = {
     isAmbiguous: true,
     name: "J",
     aliases: "IL",
-    mass: 0,
+    mass: 0
   },
   X: {
     value: "X",
@@ -240,7 +245,7 @@ const proteinAlphabet = {
     isAmbiguous: true,
     name: "X",
     aliases: "ACDEFGHIKLMNPQRSTVWY",
-    mass: 0,
+    mass: 0
   },
   Z: {
     value: "Z",
@@ -254,4 +259,4 @@ const proteinAlphabet = {
   }
 };
 
-export default proteinAlphabet
+export default proteinAlphabet;

package/src/rotateBpsToPosition.js

@@ -1,13 +1,9 @@
-export default function rotateBpsToPosition(
-  bps,
-  caretPosition,
-) {
-  return arrayRotate(bps.split(""), caretPosition).join("")
-};
-
+export default function rotateBpsToPosition(bps, caretPosition) {
+  return arrayRotate(bps.split(""), caretPosition).join("");
+}
 
 function arrayRotate(arr, count) {
-  count -= arr.length * Math.floor(count / arr.length)
-  arr.push.apply(arr, arr.splice(0, count))
-  return arr
-}
+  count -= arr.length * Math.floor(count / arr.length);
+  arr.push.apply(arr, arr.splice(0, count));
+  return arr;
+}

package/src/rotateBpsToPosition.test.js

@@ -1,6 +1,6 @@
-import rotateBpsToPosition from './rotateBpsToPosition';
-describe('rotateBpsToPosition', () => {
-    it('should rotate Bps To Position correctly ', () => {
-        expect(rotateBpsToPosition("atgaccc",4)).toEqual("cccatga")
-    })
-})
+import rotateBpsToPosition from "./rotateBpsToPosition";
+describe("rotateBpsToPosition", () => {
+  it("should rotate Bps To Position correctly ", () => {
+    expect(rotateBpsToPosition("atgaccc", 4)).toEqual("cccatga");
+  });
+});

package/src/rotateSequenceDataToPosition.js

@@ -1,7 +1,7 @@
-import {map} from "lodash";
-import {adjustRangeToRotation} from "@teselagen/range-utils";
+import { map } from "lodash";
+import { adjustRangeToRotation } from "@teselagen/range-utils";
 import tidyUpSequenceData from "./tidyUpSequenceData";
-import {modifiableTypes} from "./annotationTypes";
+import { modifiableTypes } from "./annotationTypes";
 import rotateBpsToPosition from "./rotateBpsToPosition";
 
 export default function rotateSequenceDataToPosition(
@@ -28,7 +28,7 @@ export default function rotateSequenceDataToPosition(
     );
   });
   return newSequenceData;
-};
+}
 
 function adjustAnnotationsToRotation(
   annotationsToBeAdjusted,

package/src/shiftAnnotationsByLen.js

@@ -1,4 +1,4 @@
-import {modifiableTypes} from "./annotationTypes";
+import { modifiableTypes } from "./annotationTypes";
 import adjustAnnotationsToInsert from "./adjustAnnotationsToInsert";
 
 export default function shiftAnnotationsByLen({
@@ -14,4 +14,4 @@ export default function shiftAnnotationsByLen({
       insertLength
     );
   });
-};
+}

package/src/tidyUpAnnotation.js

@@ -1,5 +1,5 @@
-import {cloneDeep, get, some} from "lodash";
-import {getFeatureToColorMap, getFeatureTypes} from "./featureTypesAndColors";
+import { cloneDeep, get, some } from "lodash";
+import { getFeatureToColorMap, getFeatureTypes } from "./featureTypesAndColors";
 import shortid from "shortid";
 
 export default function tidyUpAnnotation(
@@ -114,8 +114,9 @@ export default function tidyUpAnnotation(
       annotation.notes = JSON.parse(annotation.notes);
     } catch (error) {
       console.info(
-        `warning 33y00a0912 - couldn't parse notes for ${annotation.name ||
-          ""} ${annotation.notes}:`,
+        `warning 33y00a0912 - couldn't parse notes for ${
+          annotation.name || ""
+        } ${annotation.notes}:`,
         error
       );
     }
@@ -125,7 +126,7 @@ export default function tidyUpAnnotation(
     annotation.color = getFeatureToColorMap()[annotation.type];
   }
   return annotation;
-};
+}
 
 function coerceLocation({
   location,
@@ -156,7 +157,7 @@ function coerceLocation({
         " and set to size: " +
         size
     ); //setting it to 0 internally, but users will see it as 1
-    location.start = size - (isProtein ? 3 : 1);
+    location.start = Math.max(0, size - (isProtein ? 3 : 1));
   }
   if (
     location.end < 0 ||
@@ -171,7 +172,7 @@ function coerceLocation({
         " and set to seq size: " +
         size
     ); //setting it to 0 internally, but users will see it as 1
-    location.end = size - 1;
+    location.end = Math.max(0, size - 1);
   }
   if (location.start > location.end && circular === false) {
     messages.push(

package/src/tidyUpSequenceData.js

@@ -2,13 +2,12 @@
 import shortid from "shortid";
 
 import getAminoAcidDataForEachBaseOfDna from "./getAminoAcidDataForEachBaseOfDna";
-import {cloneDeep, flatMap} from "lodash";
+import { cloneDeep, flatMap } from "lodash";
 import { annotationTypes } from "./annotationTypes";
 import filterSequenceString from "./filterSequenceString";
 import tidyUpAnnotation from "./tidyUpAnnotation";
-import filterAminoAcidSequenceString from "./filterAminoAcidSequenceString";
 import getDegenerateDnaStringFromAaString from "./getDegenerateDnaStringFromAAString";
-import {getFeatureTypes} from "./featureTypesAndColors";
+import { getFeatureTypes } from "./featureTypesAndColors";
 
 export default function tidyUpSequenceData(pSeqData, options = {}) {
   const {
@@ -17,11 +16,10 @@ export default function tidyUpSequenceData(pSeqData, options = {}) {
     removeUnwantedChars,
     additionalValidChars,
     noTranslationData,
-    charOverrides,
     doNotProvideIdsForAnnotations,
-    proteinFilterOptions,
     noCdsTranslations,
-    convertAnnotationsFromAAIndices
+    convertAnnotationsFromAAIndices,
+    topLevelSeqData
   } = options;
   let seqData = cloneDeep(pSeqData); //sequence is usually immutable, so we clone it and return it
   const response = {
@@ -56,18 +54,17 @@ export default function tidyUpSequenceData(pSeqData, options = {}) {
   }
   if (removeUnwantedChars) {
     if (seqData.isProtein) {
-      seqData.proteinSequence = filterAminoAcidSequenceString(
-        seqData.proteinSequence,
-        { includeStopCodon: true, ...proteinFilterOptions }
-      );
+      const [newSeq] = filterSequenceString(seqData.proteinSequence, {
+        includeStopCodon: true,
+        ...(topLevelSeqData || seqData)
+      });
+      seqData.proteinSequence = newSeq;
     } else {
-      seqData.sequence = filterSequenceString(
-        seqData.sequence,
-        `${additionalValidChars || ""}${
-          seqData.isRna || seqData.isMixedRnaAndDna ? "u" : "" //if it is rna or mixed, allow u's
-        }`,
-        charOverrides
-      );
+      const [newSeq] = filterSequenceString(seqData.sequence, {
+        additionalValidChars,
+        ...(topLevelSeqData || seqData)
+      });
+      seqData.sequence = newSeq;
     }
   }
   if (seqData.isProtein) {
@@ -166,4 +163,4 @@ export default function tidyUpSequenceData(pSeqData, options = {}) {
     console.info("tidyUpSequenceData messages:", response.messages);
   }
   return seqData;
-};
+}

package/filterAminoAcidSequenceString.d.ts

@@ -1 +0,0 @@
-export default function filterAminoAcidSequenceString(sequenceString: any, options: any): any;

package/filterAminoAcidSequenceString.test.d.ts

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

package/src/filterAminoAcidSequenceString.js

@@ -1,13 +0,0 @@
-//
-export default function filterAminoAcidSequenceString(
-  sequenceString,
-  options
-) {
-  options = options || {};
-  if (options.includeStopCodon) {
-    //tnrtodo this maybe needs the stop codon char in it?
-    return sequenceString.replace(/[^xtgalmfwkqespvicyhrndu.*]/gi, "");
-  }
-  // ac.throw(ac.string, sequenceString);
-  return sequenceString.replace(/[^xtgalmfwkqespvicyhrndu]/gi, "");
-};

package/src/filterAminoAcidSequenceString.test.js

@@ -1,22 +0,0 @@
-import assert from "assert";
-import filterAminoAcidSequenceString from "./filterAminoAcidSequenceString";
-describe("filterAminoAcidSequenceString", () => {
-  it("should filter only valid amino acids by default", () => {
-    const filteredString = filterAminoAcidSequenceString(
-      'bbb342"""xtgalmfwkqespvicyhrnd,,../'
-    );
-    assert.equal(filteredString, "xtgalmfwkqespvicyhrnd");
-  });
-  it("should handle upper case letters", () => {
-    const filteredString = filterAminoAcidSequenceString("xtgalmfWKQEspvicyhrnd");
-    assert.equal(filteredString, "xtgalmfWKQEspvicyhrnd");
-  });
-  it("should handle the option to includeStopCodon by allowing periods", () => {
-    const options = { includeStopCodon: true };
-    const filteredString = filterAminoAcidSequenceString(
-      'bbb342"""xtgalmfwkqespvicyhrnd,,../',
-      options
-    );
-    assert.equal(filteredString, "xtgalmfwkqespvicyhrnd..");
-  });
-});