updating getVirtualDigest to take a couple more options

TeselaGen · Feb 27, 2023 · 045a8e6 · 045a8e6
1 parent e587961
commit 045a8e6
Show file tree

Hide file tree

Showing 8 changed files with 1,862 additions and 1,132 deletions.
diff --git a/package.json b/package.json
@@ -43,16 +43,16 @@
     "string-splice": "^1.3.0"
   },
   "devDependencies": {
-    "@types/jest": "^23.3.11",
-    "auto-changelog": "^2.2.1",
+    "@types/jest": "^29.4.0",
+    "auto-changelog": "^2.4.0",
     "babel-preset-es2015": "^6.24.1",
     "babel-preset-react": "^6.24.1",
-    "chai": "^3.2.0",
+    "chai": "^4.3.7",
     "chai-subset": "^1.1.0",
-    "eslint": "6",
-    "eslint-config-teselagen": "6.0.12",
+    "eslint": "8.35.0",
+    "eslint-config-teselagen": "6.0.16",
     "husky": "^0.11.6",
-    "jest": "^23.6.0",
+    "jest": "^29.4.3",
     "jsonfile": "^2.3.1",
     "lint-staged": "^9.2.1",
     "lodash": "^4.17.11",

diff --git a/src/cutSequenceByRestrictionEnzyme.test.js b/src/cutSequenceByRestrictionEnzyme.test.js
@@ -24,7 +24,7 @@ describe("a simple, palindromic enzyme", function() {
       true,
       enzymeList["sapi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(0);
     cutsites[0].end.should.equal(5);
@@ -42,7 +42,7 @@ describe("a simple, palindromic enzyme", function() {
       true,
       enzymeList["bamhi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(0);
     cutsites[0].end.should.equal(5);
@@ -60,7 +60,7 @@ describe("a simple, palindromic enzyme", function() {
       true,
       enzymeList["bamhi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(6);
     cutsites[0].end.should.equal(1);
@@ -77,7 +77,7 @@ describe("a simple, palindromic enzyme", function() {
       false,
       enzymeList["bamhi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(0);
   });
   it("cuts multiple times", function() {
@@ -95,7 +95,7 @@ describe("a simple, palindromic enzyme", function() {
       true,
       enzymeList["bamhi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(2);
     cutsites[0].start.should.equal(0);
     cutsites[0].end.should.equal(5);
@@ -129,7 +129,7 @@ describe("a simple, palindromic enzyme", function() {
         bottomSnipOffset: 1
       }
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(0);
     cutsites.error.should.not.be.null;
     cutsites.error.should.equal(
@@ -153,7 +153,7 @@ describe("non-palindromic enzyme", function() {
       false,
       enzymeList["bsmbi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(0);
   });
   it("does cut if the enzyme fits within circular sequence", function() {
@@ -162,7 +162,7 @@ describe("non-palindromic enzyme", function() {
       true,
       enzymeList["bsmbi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(0);
     cutsites[0].end.should.equal(4);
@@ -183,7 +183,7 @@ describe("non-palindromic enzyme", function() {
       true,
       enzymeList["bsmbi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(0);
     cutsites[0].end.should.equal(10);
@@ -205,7 +205,7 @@ describe("non-palindromic enzyme", function() {
       true,
       enzymeList["bsmbi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(18);
     cutsites[0].end.should.equal(28);
@@ -233,7 +233,7 @@ describe("palindromic enzyme that cuts both upstream and downstream", function()
       false,
       enzymeList["nmedi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(0);
   });
   it("does cut twice if the enzyme fits within linear sequence", function() {
@@ -244,7 +244,7 @@ describe("palindromic enzyme that cuts both upstream and downstream", function()
       false,
       enzymeList["nmedi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(8);
     cutsites[0].end.should.equal(37);
@@ -263,7 +263,7 @@ describe("palindromic enzyme that cuts both upstream and downstream", function()
       false,
       enzymeList["nmedi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(8);
     cutsites[0].end.should.equal(25);
@@ -282,7 +282,7 @@ describe("palindromic enzyme that cuts both upstream and downstream", function()
       false,
       enzymeList["nmedi"]
     );
-    cutsites.should.be.an.array;
+    cutsites.should.be.an("array");
     cutsites.length.should.equal(1);
     cutsites[0].start.should.equal(0);
     cutsites[0].end.should.equal(17);

diff --git a/src/getDigestFragmentsForRestrictionEnzymes.test.js b/src/getDigestFragmentsForRestrictionEnzymes.test.js
@@ -23,7 +23,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       false,
       enzymeList["bsai"]
     );
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(0);
   });
   it("returns 0 digestFragments for a circular seq with no cutsites", function() {
@@ -32,7 +32,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       false,
       enzymeList["bsai"]
     );
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(0);
   });
   it("gets digestFragments for a single type 2s cutsite cutting in a linear sequence", function() {
@@ -41,7 +41,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       false,
       enzymeList["bsai"]
     );
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(2);
     digestFragments[0].start.should.equal(0);
     digestFragments[0].end.should.equal(20);
@@ -75,7 +75,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
     // G G A T C C
     // C C T A G G
     //          ^
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(1);
     digestFragments[0].start.should.equal(1);
     digestFragments[0].end.should.equal(0);
@@ -95,7 +95,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       false,
       enzymeList["bamhi"]
     );
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(2);
     digestFragments[0].cut1.overhangSize.should.equal(0);
     digestFragments[0].cut1.isOverhangIncludedInFragmentSize.should.equal(
@@ -123,7 +123,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       false,
       enzymeList["bamhi"]
     );
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(3);
     digestFragments[0].cut1.type.should.equal("START_OR_END_OF_SEQ");
     digestFragments[2].cut2.type.should.equal("START_OR_END_OF_SEQ");
@@ -140,7 +140,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       true,
       enzymeList["bamhi"]
     );
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(2);
     digestFragments[0].start.should.equal(1);
     digestFragments[0].end.should.equal(13);
@@ -160,7 +160,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       true,
       enzymeList["bsai"]
     );
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(2);
     digestFragments[0].cut1.overhangSize.should.equal(4);
     digestFragments[0].cut1.isOverhangIncludedInFragmentSize.should.equal(true);
@@ -190,7 +190,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
       { computePartialDigests: true }
     );
 
-    digestFragments.should.be.an.array;
+    digestFragments.should.be.an("array");
     digestFragments.length.should.equal(3);
     digestFragments[0].start.should.equal(1);
     digestFragments[0].end.should.equal(13);
@@ -212,7 +212,7 @@ describe("getDigestFragmentsForRestrictionEnzymes", function() {
   //       enzymeList["bamhi"],
   //       { computePartialDigests: true }
   //     );
-  //     digestFragments.should.be.an.array;
+  //     digestFragments.should.be.an("array");
   //     digestFragments.length.should.equal(9);
   //     digestFragments[0].start.should.equal(1);
   //     digestFragments[0].end.should.equal(13);

diff --git a/src/getLeftAndRightOfSequenceInRangeGivenPosition.test.js b/src/getLeftAndRightOfSequenceInRangeGivenPosition.test.js
@@ -66,7 +66,7 @@ describe("getLeftAndRightOfSequenceInRangeGivenPosition", function() {
     result.rightHandSide.should.equal("gatcc");
     //
     // //
-    // cutsites.should.be.an.array;
+    // cutsites.should.be.an("array");
     // cutsites.length.should.equal(1);
     // cutsites[0].start.should.equal(6);
     // cutsites[0].end.should.equal(1);

diff --git a/src/getPossiblePartsFromSequenceAndEnzymes.test.js b/src/getPossiblePartsFromSequenceAndEnzymes.test.js
@@ -27,7 +27,7 @@ describe("getPossiblePartsFromSequenceAndEnzymes", function() {
       enzymeList["bamhi"]
     ]);
     // eslint-disable-next-line no-unused-expressions
-    parts.should.be.an.array;
+    parts.should.be.an("array");
     parts.length.should.equal(2);
     parts[0].start.should.equal(29);
     parts[0].end.should.equal(76);

diff --git a/src/getVirtualDigest.js b/src/getVirtualDigest.js
@@ -10,21 +10,28 @@ module.exports = function getVirtualDigest({
   cutsites,
   sequenceLength,
   isCircular,
-  allowPartialDigests
+  computePartialDigest,
+  computePartialDigestDisabled,
+  computeDigestDisabled
 }) {
   let fragments = [];
   const overlappingEnzymes = [];
-  let pairs = [];
+  const pairs = [];
+
   const sortedCutsites = cutsites.sort((a, b) => {
     return a.topSnipPosition - b.topSnipPosition;
   });
 
   sortedCutsites.forEach((cutsite1, index) => {
-    if (allowPartialDigests) {
-      sortedCutsites.forEach((cutsite2, index2) => {
-        pairs.push([cutsite1, cutsite2]);
+    if (computePartialDigest && !computePartialDigestDisabled) {
+      sortedCutsites.forEach((cs, index2) => {
+        // if (index2 === index + 1 || index2 === 0) { //tnw: not sure if this is necessary or not. commenting out for now
+        //   return;
+        // }
+        pairs.push([cutsite1, sortedCutsites[index2]]);
       });
-    } else {
+    }
+    if (!computeDigestDisabled) {
       pairs.push([
         cutsite1,
         sortedCutsites[index + 1]
@@ -33,9 +40,6 @@ module.exports = function getVirtualDigest({
       ]);
     }
   });
-  // pairs = uniqBy(pairs, ([cut1,cut2]) => {
-  //   return cut1.topSnipPosition > cut2.topSnipPosition ? (cut1.name || "cut1") + "_" + (cut2.name || "cut2") :  (cut2.name || "cut2") + "_" + (cut1.name || "cut1")
-  // })
 
   pairs.forEach(([cut1, cut2]) => {
     const start = normalizePositionByRangeLength(
@@ -46,61 +50,70 @@ module.exports = function getVirtualDigest({
       cut2.topSnipPosition - 1,
       sequenceLength
     );
+
     if (!isCircular && start > end) {
       //we have a fragment that spans the origin so we need to split it in 2 pieces
       const frag1 = {
         start: start,
         end: sequenceLength - 1,
         cut1,
-        cut2: "endOfSeq"
+        cut2: {
+          type: "endOfSeq",
+          restrictionEnzyme: {
+            name: "End Of Seq"
+          }
+        }
       };
       const frag2 = {
         start: 0,
         end: end,
-        cut1: "startOfSeq",
+        cut1: {
+          type: "startOfSeq",
+          restrictionEnzyme: {
+            name: "Start Of Seq"
+          }
+        },
         cut2: cut2
       };
 
-      fragments.push(addSizeAndId(frag1, sequenceLength));
-      fragments.push(addSizeAndId(frag2, sequenceLength));
+      fragments.push(addSizeIdName(frag1, sequenceLength));
+      fragments.push(addSizeIdName(frag2, sequenceLength));
     } else {
       const frag = {
         cut1,
         cut2,
         start,
         end
       };
-      fragments.push(addSizeAndId(frag, sequenceLength));
+      fragments.push(addSizeIdName(frag, sequenceLength));
     }
   });
-  // const sizeMap = {};
   fragments = fragments.filter(fragment => {
     if (!fragment.size) {
       overlappingEnzymes.push(fragment);
       return false;
     }
-    // if (sizeMap[fragment.size]) {
-    //   sizeMap[fragment.size].push(fragment);
-    //   return false;
-    // } else {
-    //   sizeMap[fragment.size] = [fragment];
-    // }
     return true;
   });
   return {
+    computePartialDigestDisabled,
+    computeDigestDisabled,
     fragments,
     overlappingEnzymes
   };
 };
 
-function addSizeAndId(frag, sequenceLength) {
+function addSizeIdName(frag, sequenceLength) {
   const size = getRangeLength(
     { start: frag.start, end: frag.end },
     sequenceLength
   );
+  const name = `${frag.cut1.restrictionEnzyme?.name} -- ${frag.cut2.restrictionEnzyme?.name} ${size} bps`;
+
   return {
     ...frag,
     size,
+    name,
     id: frag.start + "-" + frag.end + "-" + size + "-"
   };
 }