@teselagen/sequence-utils 0.1.21 → 0.1.23

package/package.json

@@ -1,9 +1,9 @@
 {
   "name": "@teselagen/sequence-utils",
-  "version": "0.1.21",
+  "version": "0.1.23",
   "type": "commonjs",
   "dependencies": {
-    "@teselagen/range-utils": "0.1.20",
+    "@teselagen/range-utils": "0.1.22",
     "bson-objectid": "^2.0.4",
     "escape-string-regexp": "^5.0.0",
     "jsondiffpatch-rc": "0.4.2",

package/src/DNAComplementMap.js

@@ -0,0 +1,32 @@
+const DNAComplementMap = {
+  ".": ".",
+  a: "t",
+  t: "a",
+  u: "a",
+  c: "g",
+  g: "c",
+  A: "T",
+  T: "A",
+  U: "A",
+  C: "G",
+  G: "C",
+  r: "y",
+  R: "Y",
+  y: "r",
+  Y: "R",
+  d: "h",
+  D: "H",
+  h: "d",
+  H: "D",
+  k: "m",
+  K: "M",
+  m: "k",
+  M: "K",
+  v: "b",
+  V: "B",
+  b: "v",
+  B: "V"
+  //tnrtodo add more letters here
+};
+
+export default DNAComplementMap;

package/src/addGapsToSeqReads.js

@@ -0,0 +1,417 @@
+import insertGapsIntoRefSeq from "./insertGapsIntoRefSeq.js";
+
+import {cloneDeep} from "lodash";
+
+// bam.seq: NTGTAAGTCGTGAAAAAANCNNNCATATTNCGGAGGTAAAAATGAAAA...
+// bam.pos: 43
+// bam.cigar: 36M2D917M3I17M7I2M1I6M5I4M1D6M12I8M
+// (note: bam.cigar is null if the sequencing read is unaligned)
+// bam.reversed: true (if reversed)
+
+// refSeq should be an object { name, sequence }
+// seqReads should be an array of objects [{name, seq, pos, cigar}, {name, seq, pos, cigar}, ...]
+// add gaps into sequencing reads before starting bp pos and from own deletions & all seq reads' insertions, minus own insertions
+export default function addGapsToSeqReads(refSeq, seqReads) {
+  // remove unaligned seq reads for now
+  for (let i = 0; i < seqReads.length; i++) {
+    if (seqReads[i].cigar === null) {
+      seqReads.splice(i, 1);
+    }
+  }
+
+  const refSeqWithGaps = insertGapsIntoRefSeq(refSeq.sequence, seqReads);
+  // first object is reference sequence with gaps, to be followed by seq reads with gaps
+  const seqReadsWithGaps = [
+    { name: refSeq.name, sequence: refSeqWithGaps.toUpperCase() }
+  ];
+  seqReads.forEach(seqRead => {
+    // get all insertions in seq reads
+    const allInsertionsInSeqReads = [];
+    seqReads.forEach(seqRead => {
+      // split cigar string at S, M, D, or I (soft-clipped, match, deletion, or insertion), e.g. ["5S", "2M", "3I", "39M", "3D"..."9S"]
+      const splitSeqRead = seqRead.cigar.match(/([0-9]*[SMDI])/g);
+      // adjust seqRead.pos, aka bp pos where the seq read starts aligning to the ref seq, if bps have been soft-clipped from the beginning of the seq read
+      let adjustedSeqReadPos = cloneDeep(seqRead.pos);
+      if (splitSeqRead[0].slice(-1) === "S") {
+        // # in #S at beginning of array, i.e. number of soft-clipped base pairs at beginning of the seq read
+        const numOfBeginningSoftClipped = splitSeqRead[0].slice(0, -1);
+        adjustedSeqReadPos = seqRead.pos - numOfBeginningSoftClipped;
+      }
+      for (let componentI = 0; componentI < splitSeqRead.length; componentI++) {
+        if (splitSeqRead[componentI].slice(-1) === "I") {
+          let bpPosOfInsertion = adjustedSeqReadPos;
+          const numberOfInsertions = Number(
+            splitSeqRead[componentI].slice(0, -1)
+          );
+          for (let i = 0; i < componentI; i++) {
+            if (splitSeqRead[i].slice(-1) !== "I") {
+              const previousComponentNumber = Number(
+                splitSeqRead[i].slice(0, -1)
+              );
+              bpPosOfInsertion += previousComponentNumber;
+            }
+          }
+          const insertionInfo = {
+            // keeping bpPos 1-based
+            bpPos: bpPosOfInsertion,
+            number: numberOfInsertions
+          };
+          allInsertionsInSeqReads.push(insertionInfo);
+        }
+      }
+    });
+
+    // 1) add gaps before starting bp pos
+    const splitSeqReadChunk = seqRead.cigar.match(/([0-9]*[SMDI])/g);
+    let adjustedSeqReadPos = cloneDeep(seqRead.pos);
+    if (splitSeqReadChunk[0].slice(-1) === "S") {
+      // # in #S at beginning of array, i.e. number of soft-clipped base pairs at beginning of the seq read
+      const numOfBeginningSoftClipped = splitSeqReadChunk[0].slice(0, -1);
+      adjustedSeqReadPos = seqRead.pos - numOfBeginningSoftClipped;
+    }
+    let eachSeqReadWithGaps = seqRead.seq.split("");
+    if (adjustedSeqReadPos > 0) {
+      eachSeqReadWithGaps.unshift("-".repeat(adjustedSeqReadPos - 1));
+    }
+    eachSeqReadWithGaps = eachSeqReadWithGaps.join("").split("");
+
+    // 2) add own deletions to own sequence
+    // get own deletions
+    const ownDeletions = [];
+    for (
+      let componentI = 0;
+      componentI < splitSeqReadChunk.length;
+      componentI++
+    ) {
+      if (splitSeqReadChunk[componentI].slice(-1) === "D") {
+        let bpPosOfDeletion = adjustedSeqReadPos;
+        const numberOfDeletions = Number(
+          splitSeqReadChunk[componentI].slice(0, -1)
+        );
+        for (let i = 0; i < componentI; i++) {
+          const previousComponentNumber = Number(
+            splitSeqReadChunk[i].slice(0, -1)
+          );
+          bpPosOfDeletion += previousComponentNumber;
+        }
+        const deletionInfo = {
+          // keeping bpPos 1-based
+          bpPos: bpPosOfDeletion,
+          number: numberOfDeletions
+        };
+        ownDeletions.push(deletionInfo);
+      }
+    }
+    // sort deletions by ascending bp pos
+    const sortedOwnDeletions = ownDeletions.sort((a, b) => {
+      return a.bpPos - b.bpPos;
+    });
+    // add own deletions to own sequence
+    for (let ownD = 0; ownD < sortedOwnDeletions.length; ownD++) {
+      const bpPosOfDeletion = sortedOwnDeletions[ownD].bpPos;
+      const numberOfDeletions = sortedOwnDeletions[ownD].number;
+      // adding gaps at the bp pos
+      let deletionGaps = "";
+      for (let gapD = 0; gapD < numberOfDeletions; gapD++) {
+        deletionGaps += "-";
+      }
+      eachSeqReadWithGaps.splice(bpPosOfDeletion - 1, 0, deletionGaps);
+      eachSeqReadWithGaps = eachSeqReadWithGaps.join("").split("");
+    }
+    eachSeqReadWithGaps = eachSeqReadWithGaps.join("").split("");
+
+    // 3) remove own insertions from own sequence
+    // get own insertions
+    const ownInsertions = [];
+    const ownInsertionsBp = [];
+    for (
+      let componentI = 0;
+      componentI < splitSeqReadChunk.length;
+      componentI++
+    ) {
+      if (splitSeqReadChunk[componentI].slice(-1) === "I") {
+        let bpPosOfInsertion = adjustedSeqReadPos;
+        const numberOfInsertions = Number(
+          splitSeqReadChunk[componentI].slice(0, -1)
+        );
+        const nucleotides = [];
+        for (let i = 0; i < componentI; i++) {
+          const previousComponentNumber = Number(
+            splitSeqReadChunk[i].slice(0, -1)
+          );
+          bpPosOfInsertion += previousComponentNumber;
+        }
+        for (let nucI = 0; nucI < numberOfInsertions; nucI++) {
+          nucleotides.push(eachSeqReadWithGaps[bpPosOfInsertion - 1 + nucI]);
+        }
+        const insertionInfo = {
+          // keeping bpPos 1-based
+          bpPos: bpPosOfInsertion,
+          number: numberOfInsertions
+        };
+        const insertionInfoBp = {
+          // keeping bpPos 1-based
+          bpPos: bpPosOfInsertion,
+          number: numberOfInsertions,
+          nucleotides: nucleotides
+        };
+        ownInsertions.push(insertionInfo);
+        ownInsertionsBp.push(insertionInfoBp);
+      }
+    }
+    const ownInsertionsCompare = JSON.parse(JSON.stringify(ownInsertions));
+    // sort own insertions by ascending bp pos
+    const sortedOwnInsertions = ownInsertions.sort((a, b) => {
+      return a.bpPos - b.bpPos;
+    });
+    const sortedOwnInsertionsBp = ownInsertionsBp.sort((a, b) => {
+      return a.bpPos - b.bpPos;
+    });
+    // remove own insertions from own sequence
+    for (let ownI = 0; ownI < sortedOwnInsertions.length; ownI++) {
+      const bpPosOfInsertion = sortedOwnInsertions[ownI].bpPos;
+      const numberOfInsertions = sortedOwnInsertions[ownI].number;
+      for (let numI = 0; numI < numberOfInsertions; numI++) {
+        eachSeqReadWithGaps.splice(bpPosOfInsertion - 1, 1);
+      }
+      for (let posI = ownI + 1; posI < sortedOwnInsertions.length; posI++) {
+        sortedOwnInsertions[posI].bpPos -= numberOfInsertions;
+      }
+    }
+
+    // 4) add other seq reads' insertions to seq read
+    // get other seq reads' insertions (i.e. all insertions minus duplicates minus own insertions)
+    let otherInsertions = allInsertionsInSeqReads.sort((a, b) => {
+      return a.bpPos - b.bpPos;
+    });
+    // combine duplicates within all insertions, remove own insertions from all insertions, combine overlap between other insertions & own insertions
+    // first, combine duplicates within all insertions
+    otherInsertions = otherInsertions.filter(
+      (object, index) =>
+        index ===
+        otherInsertions.findIndex(
+          obj => JSON.stringify(obj) === JSON.stringify(object)
+        )
+    );
+    // 'i < otherInsertions.length - 1' because when at the end of the array, there is no 'i + 1' to compare to
+    for (let i = 0; i < otherInsertions.length - 1; i++) {
+      while (otherInsertions[i].bpPos === otherInsertions[i + 1].bpPos) {
+        if (otherInsertions[i].number > otherInsertions[i + 1].number) {
+          // remove the one with fewer number of gaps from array
+          otherInsertions.splice(i + 1, 1);
+        } else if (otherInsertions[i].number < otherInsertions[i + 1].number) {
+          otherInsertions.splice(i, 1);
+        } else if (
+          otherInsertions[i].number === otherInsertions[i + 1].number
+        ) {
+          otherInsertions.splice(i, 1);
+        }
+      }
+    }
+    // then remove own insertions from all insertions
+    for (let otherI = 0; otherI < ownInsertionsCompare.length; otherI++) {
+      const insertionInfoIndex = otherInsertions.findIndex(
+        e => e.bpPos === ownInsertionsCompare[otherI].bpPos
+      );
+      if (insertionInfoIndex !== -1) {
+        if (
+          otherInsertions[insertionInfoIndex].number >
+          ownInsertionsCompare[otherI].number
+        ) {
+          otherInsertions[insertionInfoIndex].number =
+            otherInsertions[insertionInfoIndex].number -
+            ownInsertionsCompare[otherI].number;
+        } else if (
+          otherInsertions[insertionInfoIndex].number <=
+          ownInsertionsCompare[otherI].number
+        ) {
+          otherInsertions.splice(insertionInfoIndex, 1);
+          otherI--;
+        }
+      }
+    }
+    // then combine overlap between other insertions & own insertions
+    for (let overlapI = 0; overlapI < sortedOwnInsertions.length; overlapI++) {
+      const insertionInfoIndex = otherInsertions.findIndex(
+        e => e.bpPos === sortedOwnInsertions[overlapI].bpPos
+      );
+      if (insertionInfoIndex !== -1) {
+        if (
+          otherInsertions[insertionInfoIndex].number >
+          sortedOwnInsertions[overlapI].number
+        ) {
+          otherInsertions[insertionInfoIndex].number =
+            otherInsertions[insertionInfoIndex].number -
+            sortedOwnInsertions[overlapI].number;
+        } else if (
+          otherInsertions[insertionInfoIndex].number <=
+          sortedOwnInsertions[overlapI].number
+        ) {
+          otherInsertions.splice(insertionInfoIndex, 1);
+          overlapI--;
+        }
+      }
+    }
+    // adjust own insertions according to other seq reads' insertions to be added (i.e. for all other reads' insertions with smaller bp pos, +1 to that own insertion's bp pos)
+    const adjustedOwnInsertionsBp = JSON.parse(
+      JSON.stringify(sortedOwnInsertionsBp)
+    );
+    for (let ownI = 0; ownI < adjustedOwnInsertionsBp.length; ownI++) {
+      let previousInserts = 0;
+      for (let i = 0; i < ownI; i++) {
+        previousInserts += adjustedOwnInsertionsBp[i].number - 1;
+      }
+      adjustedOwnInsertionsBp[ownI].bpPos =
+        adjustedOwnInsertionsBp[ownI].bpPos - previousInserts;
+      sortedOwnInsertionsBp[ownI].bpPos =
+        sortedOwnInsertionsBp[ownI].bpPos - previousInserts;
+    }
+    for (let otherI = 0; otherI < otherInsertions.length; otherI++) {
+      for (let ownI = 0; ownI < adjustedOwnInsertionsBp.length; ownI++) {
+        if (
+          otherInsertions[otherI].bpPos <= sortedOwnInsertionsBp[ownI].bpPos
+        ) {
+          adjustedOwnInsertionsBp[ownI].bpPos += 1;
+        }
+      }
+    }
+    // add other seq reads' insertions to sequence
+    for (
+      let otherI = 0;
+      otherI < otherInsertions.length &&
+      otherInsertions[otherI].bpPos <= eachSeqReadWithGaps.length;
+      otherI++
+    ) {
+      const bpPosOfInsertion = otherInsertions[otherI].bpPos;
+      const numberOfInsertions = otherInsertions[otherI].number;
+      // adding gaps at the bp pos
+      let insertionGaps = "";
+      for (let gapI = 0; gapI < numberOfInsertions; gapI++) {
+        insertionGaps += "-";
+      }
+      eachSeqReadWithGaps.splice(bpPosOfInsertion - 1, 0, insertionGaps);
+      for (let posI = otherI + 1; posI < otherInsertions.length; posI++) {
+        otherInsertions[posI].bpPos += 1;
+      }
+    }
+
+    // 5) add own insertions to own sequence
+    for (let ownI = 0; ownI < adjustedOwnInsertionsBp.length; ownI++) {
+      const bpPosOfInsertion = adjustedOwnInsertionsBp[ownI].bpPos;
+      const nucleotides = adjustedOwnInsertionsBp[ownI].nucleotides.join("");
+      eachSeqReadWithGaps.splice(bpPosOfInsertion - 1, 0, nucleotides);
+    }
+
+    // 6) add gaps after seq read for ref seq's length = seq read's length
+    eachSeqReadWithGaps = eachSeqReadWithGaps.join("").split("");
+    if (eachSeqReadWithGaps.length < refSeqWithGaps.length) {
+      eachSeqReadWithGaps.push(
+        "-".repeat(refSeqWithGaps.length - eachSeqReadWithGaps.length)
+      );
+    }
+
+    // eachSeqReadWithGaps is a string "GGGA--GA-C--ACC"
+    seqReadsWithGaps.push({
+      name: seqRead.name,
+      sequence: eachSeqReadWithGaps.join(""),
+      reversed: seqRead.reversed,
+      cigar: seqRead.cigar
+    });
+  });
+
+  // 7) add gaps before starting bp pos
+  // add gaps based on any seq reads that extend beyond beginning of the ref seq due to soft-clipped reads
+  // a) get the lengths of bps that extend beyond the beginning of the ref seq among all seq reads
+  const seqReadLengthsBeforeRefSeqStart = [];
+  seqReads.forEach(seq => {
+    const splitSeqReadChunk = seq.cigar.match(/([0-9]*[SMDI])/g);
+    let adjustedSeqReadPos = cloneDeep(seq.pos);
+    if (splitSeqReadChunk[0].slice(-1) === "S") {
+      // # in #S at beginning of array, i.e. number of soft-clipped base pairs at beginning of the seq read
+      const numOfBeginningSoftClipped = splitSeqReadChunk[0].slice(0, -1);
+      adjustedSeqReadPos = seq.pos - numOfBeginningSoftClipped;
+      // number of gaps to add if soft-clipped reads extend beyond beginning of ref seq
+      if (adjustedSeqReadPos < 0) {
+        seqReadLengthsBeforeRefSeqStart.push(Math.abs(adjustedSeqReadPos));
+      }
+    }
+    // number of gaps to add if seqRead.pos is negative (not sure if this is possible with bowtie2 outputs)
+    // if (seq.pos < 0) {
+    //   seqReadLengthsBeforeRefSeqStart.push(Math.abs(seq.pos))
+    // }
+  });
+  // b) add gaps (to both ref seq and seq reads) based on any seq reads that extend beyond beginning of ref seq due to soft-clipped reads
+  let longestSeqReadLength = 0;
+  for (let i = 1; i < seqReadsWithGaps.length; i++) {
+    // turn seq read into an array ["A", "T", "C", "G"...]
+    const eachSeqReadWithGaps = seqReadsWithGaps[i].sequence.split("");
+    const splitSeqReadChunk = seqReads[i - 1].cigar.match(/([0-9]*[SMDI])/g);
+    let adjustedSeqReadPos = cloneDeep(seqReads[i - 1].pos);
+    // longest length of bps that extend beyond the beginning of the ref seq among all seq reads
+    if (seqReadLengthsBeforeRefSeqStart.length > 0) {
+      longestSeqReadLength = Math.max(...seqReadLengthsBeforeRefSeqStart);
+    }
+    if (splitSeqReadChunk[0].slice(-1) === "S") {
+      // # in #S at beginning of array, i.e. number of soft-clipped base pairs at beginning of the seq read
+      const numOfBeginningSoftClipped = splitSeqReadChunk[0].slice(0, -1);
+      adjustedSeqReadPos = seqReads[i - 1].pos - numOfBeginningSoftClipped;
+      if (adjustedSeqReadPos > 0) {
+        if (longestSeqReadLength > 0) {
+          eachSeqReadWithGaps.unshift("-".repeat(longestSeqReadLength + 1));
+        }
+        seqReadsWithGaps[i].sequence = eachSeqReadWithGaps.join("");
+      } else if (adjustedSeqReadPos < 0) {
+        if (longestSeqReadLength > 0) {
+          eachSeqReadWithGaps.unshift(
+            "-".repeat(longestSeqReadLength - Math.abs(adjustedSeqReadPos))
+          );
+        }
+        seqReadsWithGaps[i].sequence = eachSeqReadWithGaps.join("");
+      }
+    } else {
+      if (longestSeqReadLength > 0) {
+        eachSeqReadWithGaps.unshift("-".repeat(longestSeqReadLength + 1));
+      }
+      seqReadsWithGaps[i].sequence = eachSeqReadWithGaps.join("");
+    }
+  }
+
+  // add gaps before ref seq based on the longest length of soft-clipped reads that extend beyond beginning of ref seq
+  if (longestSeqReadLength > 0) {
+    const splitRefSeqWithGaps = seqReadsWithGaps[0].sequence.split("");
+    splitRefSeqWithGaps.unshift("-".repeat(longestSeqReadLength + 1));
+    seqReadsWithGaps[0].sequence = splitRefSeqWithGaps.join("");
+  }
+
+  // 8) check if any seq read is longer than the ref seq, make ref seq & seq reads all the same length
+  const lengthsOfLongerSeqReads = [];
+  for (let i = 1; i < seqReadsWithGaps.length; i++) {
+    const refSeq = seqReadsWithGaps[0];
+    if (seqReadsWithGaps[i].sequence.length > refSeq.sequence.length) {
+      lengthsOfLongerSeqReads.push(seqReadsWithGaps[i].sequence.length);
+    }
+  }
+  if (lengthsOfLongerSeqReads.length > 0) {
+    const longestSeqReadLength = Math.max(...lengthsOfLongerSeqReads);
+    for (let i = 0; i < seqReadsWithGaps.length; i++) {
+      if (seqReadsWithGaps[i].sequence.length < longestSeqReadLength) {
+        seqReadsWithGaps[i].sequence += "-".repeat(
+          longestSeqReadLength - seqReadsWithGaps[i].sequence.length
+        );
+      }
+    }
+  }
+  // if any seq read shorter than ref seq, make ref seq & seq reads all the same length
+  for (let i = 1; i < seqReadsWithGaps.length; i++) {
+    const refSeq = seqReadsWithGaps[0];
+    if (seqReadsWithGaps[i].sequence.length < refSeq.sequence.length) {
+      seqReadsWithGaps[i].sequence += "-".repeat(
+        refSeq.sequence.length - seqReadsWithGaps[i].sequence.length
+      );
+    }
+  }
+
+  // seqReadsWithGaps is an array of objects containing the ref seq with gaps first and then all seq reads with gaps
+  // e.g. [{ name: "ref seq", sequence: "GG---GA--GA-C--A---CC---"}, { name: "r1", sequence: "-----GATTGA-C-----------"}...]
+  return seqReadsWithGaps;
+};