react-msaview 4.4.6 → 4.5.0

package/src/parsers/A3mMSA.test.ts

@@ -0,0 +1,164 @@
+import { describe, expect, test } from 'vitest'
+
+import A3mMSA from './A3mMSA'
+
+const exampleA3M = `>example
+ETESMKTVRIREKIKKFLGDRPRNTAEILEHINSTMRHGTTSQQLGNVLSKDKDIVKVGYIKRSGILSGGYDICEWATRNWVAEHCPEWTE
+>1
+----MRTTRLRQKIKKFLNERGeANTTEILEHVNSTMRHGTTPQQLGNVLSKDKDILKVATTKRGGALSGRYEICVWTLRP-----------
+>2
+----MDSQNLRDLIRNYLSERPRNTIEISAWLASQMDPNSCPEDVTNILEADESIVRIGTVRKSGMRLTDLPISEWASSSWVRRHE-----
+>3
+----MNSQNLRELIRNYLSERPRNTIEISTWLSSQIDPTNSPVDITSILEADDQIVRIGTVRKSGMRRSESPVSEWASNTWVKHHE-----
+>4
+--RDMDTEKVREIVRNYISERPRNTAEIAAWLNRH-DDGTGGSDVAAILESDGSFVRIGTVRTSGMTGNSPPLSEWATEKWIQHHER----
+>5
+-----RTRRLREAVLVFLEEKGnANTVEVFDYLNERFRWGATMNQVGNILAKDTRFAKVGHQ-RGQFRGSVYTVCVWALS------------
+>6
+-----RTKRLREAVRVYLAENGrSHTVDIFDHLNDRFSWGATMNQVGNILAKDNRFEKVGHVRD-FFRGARYTVCVWDLAS-----------
+`
+
+describe('A3mMSA', () => {
+  test('sniff detects A3M format', () => {
+    expect(A3mMSA.sniff(exampleA3M)).toBe(true)
+  })
+
+  test('sniff returns false for regular FASTA', () => {
+    const fasta = `>seq1
+ACDEFGHIKLMNPQRST
+>seq2
+ACDEFGHIKLMNPQRST
+`
+    expect(A3mMSA.sniff(fasta)).toBe(false)
+  })
+
+  test('sniff returns false for non-FASTA formats', () => {
+    expect(A3mMSA.sniff('# STOCKHOLM 1.0\n')).toBe(false)
+    expect(A3mMSA.sniff('CLUSTAL W')).toBe(false)
+  })
+
+  test('parses A3M and expands insertions', () => {
+    const parser = new A3mMSA(exampleA3M)
+    const names = parser.getNames()
+
+    expect(names).toEqual(['example', '1', '2', '3', '4', '5', '6'])
+
+    // All sequences should have the same length after expansion
+    const widths = names.map(name => parser.getRow(name).length)
+    expect(widths.every(w => w === widths[0])).toBe(true)
+
+    // The width should be greater than the original due to expanded inserts
+    expect(parser.getWidth()).toBeGreaterThan(90)
+  })
+
+  test('lowercase inserts become uppercase after expansion', () => {
+    const parser = new A3mMSA(exampleA3M)
+
+    // The expanded sequences should not contain lowercase letters
+    for (const name of parser.getNames()) {
+      const row = parser.getRow(name)
+      expect(/[a-z]/.test(row)).toBe(false)
+    }
+  })
+
+  test('handles simple A3M with single insert', () => {
+    // In valid A3M, match columns (uppercase + - + .) must be consistent
+    // seq1 has 5 match columns + 1 insert after D
+    // seq2 has 5 match columns, no inserts
+    const simple = `>seq1
+ACDaEF
+>seq2
+ACDEF
+`
+    const parser = new A3mMSA(simple)
+
+    // seq1 has an 'a' insert after 'D', seq2 doesn't
+    // After expansion, both should be same length
+    const seq1 = parser.getRow('seq1')
+    const seq2 = parser.getRow('seq2')
+
+    expect(seq1.length).toBe(seq2.length)
+    expect(seq1).toBe('ACDAEF')
+    expect(seq2).toBe('ACD.EF')
+  })
+
+  test('handles multiple inserts at different positions', () => {
+    // seq1: 6 match columns (A,C,D,E,F,I) + inserts (ab after D, gh after F)
+    // seq2: 6 match columns, no inserts
+    const multi = `>seq1
+ACDabEFghI
+>seq2
+ACDEFI
+`
+    const parser = new A3mMSA(multi)
+
+    const seq1 = parser.getRow('seq1')
+    const seq2 = parser.getRow('seq2')
+
+    expect(seq1.length).toBe(seq2.length)
+    // seq1: ACD + ab (inserts) + EF + gh (inserts) + I
+    // seq2: ACD + EF + I -> needs . padding at insert positions
+    expect(seq1).toBe('ACDABEFGHI')
+    expect(seq2).toBe('ACD..EF..I')
+  })
+
+  test('handles varying insert lengths', () => {
+    const varying = `>seq1
+ACDabcEF
+>seq2
+ACDaEF
+>seq3
+ACDEF
+`
+    const parser = new A3mMSA(varying)
+
+    const seq1 = parser.getRow('seq1')
+    const seq2 = parser.getRow('seq2')
+    const seq3 = parser.getRow('seq3')
+
+    // All should have same length
+    expect(seq1.length).toBe(seq2.length)
+    expect(seq2.length).toBe(seq3.length)
+
+    // seq1 has 3 inserts, seq2 has 1, seq3 has 0
+    // After expansion with max 3 insert slots:
+    expect(seq1).toBe('ACDABCEF')
+    expect(seq2).toBe('ACDA..EF')
+    expect(seq3).toBe('ACD...EF')
+  })
+
+  test('getTree returns flat tree structure', () => {
+    const parser = new A3mMSA(exampleA3M)
+    const tree = parser.getTree()
+
+    expect(tree.id).toBe('root')
+    expect(tree.noTree).toBe(true)
+    expect(tree.children.length).toBe(7)
+    expect(tree.children.map(c => c.name)).toEqual([
+      'example',
+      '1',
+      '2',
+      '3',
+      '4',
+      '5',
+      '6',
+    ])
+  })
+
+  test('getWidth returns consistent width', () => {
+    const parser = new A3mMSA(exampleA3M)
+    const width = parser.getWidth()
+
+    for (const name of parser.getNames()) {
+      expect(parser.getRow(name).length).toBe(width)
+    }
+  })
+
+  test('getMSA returns parsed data', () => {
+    const parser = new A3mMSA(exampleA3M)
+    const msa = parser.getMSA()
+
+    expect(msa.seqdata).toBeDefined()
+    expect(Object.keys(msa.seqdata).length).toBe(7)
+  })
+})

package/src/parsers/A3mMSA.ts

@@ -0,0 +1,321 @@
+import type { NodeWithIds } from '../types'
+
+// Char code helpers for fast character classification
+const CODE_A = 65 // 'A'
+const CODE_Z = 90 // 'Z'
+const CODE_a = 97 // 'a'
+const CODE_z = 122 // 'z'
+const CODE_DASH = 45 // '-'
+const CODE_DOT = 46 // '.'
+
+function isUpperOrGap(code: number): boolean {
+  return (
+    (code >= CODE_A && code <= CODE_Z) ||
+    code === CODE_DASH ||
+    code === CODE_DOT
+  )
+}
+
+function isLower(code: number): boolean {
+  return code >= CODE_a && code <= CODE_z
+}
+
+/**
+ * A3M format parser
+ *
+ * The A3M format consists of aligned fasta, in which:
+ * - Insertions are shown as lowercase characters
+ * - Matches are shown as uppercase characters
+ * - Deletions are shown as '-'
+ * - Gaps aligned to inserts are shown as '.'
+ *
+ * The key property is that lowercase letters (inserts) implicitly introduce
+ * gaps in all other sequences that don't have an insert at that position.
+ */
+export default class A3mMSA {
+  private MSA: { seqdata: Record<string, string> }
+  private orderedNames: string[]
+
+  constructor(text: string) {
+    const rawSeqs: string[] = []
+    const names: string[] = []
+
+    // First pass: parse sequences (like FASTA), preserving order
+    for (const entry of text.split('>')) {
+      if (!/\S/.test(entry)) {
+        continue
+      }
+      const newlineIdx = entry.indexOf('\n')
+      if (newlineIdx === -1) {
+        continue
+      }
+      const defLine = entry.slice(0, newlineIdx)
+      const spaceIdx = defLine.indexOf(' ')
+      const id = spaceIdx === -1 ? defLine : defLine.slice(0, spaceIdx)
+      if (id) {
+        rawSeqs.push(entry.slice(newlineIdx + 1).replaceAll(/\s/g, ''))
+        names.push(id)
+      }
+    }
+
+    this.orderedNames = names
+    this.MSA = { seqdata: this.expandA3M(rawSeqs, names) }
+  }
+
+  /**
+   * Detect if text is likely A3M format
+   */
+  static sniff(text: string): boolean {
+    if (!text.startsWith('>')) {
+      return false
+    }
+
+    const seqs: string[] = []
+    for (const entry of text.split('>')) {
+      if (!/\S/.test(entry)) {
+        continue
+      }
+      const newlineIdx = entry.indexOf('\n')
+      if (newlineIdx === -1) {
+        continue
+      }
+      const seq = entry.slice(newlineIdx + 1).replaceAll(/\s/g, '')
+      if (seq) {
+        seqs.push(seq)
+      }
+    }
+
+    if (seqs.length < 2) {
+      return false
+    }
+
+    // Check for lowercase and compute lengths in single pass per sequence
+    let hasLowercase = false
+    let firstMatchLen = -1
+    let firstRawLen = -1
+    let sameMatchLength = true
+    let differentRawLengths = false
+
+    for (const seq of seqs) {
+      let matchLen = 0
+      for (let i = 0; i < seq.length; i++) {
+        const code = seq.charCodeAt(i)
+        if (isLower(code)) {
+          hasLowercase = true
+        } else {
+          matchLen++
+        }
+      }
+
+      if (firstMatchLen === -1) {
+        firstMatchLen = matchLen
+        firstRawLen = seq.length
+      } else {
+        if (matchLen !== firstMatchLen) {
+          sameMatchLength = false
+        }
+        if (seq.length !== firstRawLen) {
+          differentRawLengths = true
+        }
+      }
+    }
+
+    return hasLowercase && sameMatchLength && differentRawLengths
+  }
+
+  /**
+   * Expand A3M format to standard aligned format.
+   *
+   * In A3M, lowercase characters are insertions that implicitly introduce
+   * gaps in sequences that don't have an insert at that position.
+   */
+  private expandA3M(
+    rawSeqs: string[],
+    names: string[],
+  ): Record<string, string> {
+    const numSeqs = names.length
+    if (numSeqs === 0) {
+      return {}
+    }
+
+    // Parse sequences into parallel arrays: matchChars and insertLengths
+    // matchChars[seqIdx] = string of match characters for that sequence
+    // insertLengths[seqIdx] = array of insert lengths after each match position
+    const matchChars: string[] = []
+    const insertLengths: number[][] = []
+
+    for (let seqIdx = 0; seqIdx < numSeqs; seqIdx++) {
+      const seq = rawSeqs[seqIdx]!
+      const matches: string[] = []
+      const insLens: number[] = []
+      let i = 0
+
+      while (i < seq.length) {
+        const code = seq.charCodeAt(i)
+
+        if (isUpperOrGap(code)) {
+          matches.push(seq[i]!)
+          // Count following lowercase inserts
+          let insLen = 0
+          let j = i + 1
+          while (j < seq.length && isLower(seq.charCodeAt(j))) {
+            insLen++
+            j++
+          }
+          insLens.push(insLen)
+          i = j
+        } else if (isLower(code)) {
+          // Leading insert before first match
+          matches.push('')
+          let insLen = 0
+          let j = i
+          while (j < seq.length && isLower(seq.charCodeAt(j))) {
+            insLen++
+            j++
+          }
+          insLens.push(insLen)
+          i = j
+        } else {
+          i++
+        }
+      }
+
+      matchChars.push(matches.join(''))
+      insertLengths.push(insLens)
+    }
+
+    // Find number of match positions and max inserts at each position
+    let numPositions = 0
+    for (let seqIdx = 0; seqIdx < numSeqs; seqIdx++) {
+      const len = insertLengths[seqIdx]!.length
+      if (len > numPositions) {
+        numPositions = len
+      }
+    }
+
+    const maxInserts = new Array<number>(numPositions).fill(0)
+    for (let seqIdx = 0; seqIdx < numSeqs; seqIdx++) {
+      const insLens = insertLengths[seqIdx]!
+      for (let pos = 0; pos < insLens.length; pos++) {
+        const len = insLens[pos]!
+        if (len > maxInserts[pos]!) {
+          maxInserts[pos] = len
+        }
+      }
+    }
+
+    // Pre-compute gap strings for common lengths (avoid repeated .repeat())
+    const gapCache: string[] = ['']
+    const maxGap = Math.max(...maxInserts, 0)
+    for (let i = 1; i <= maxGap; i++) {
+      gapCache.push('.'.repeat(i))
+    }
+
+    // Build expanded sequences
+    const expanded: Record<string, string> = {}
+
+    for (let seqIdx = 0; seqIdx < numSeqs; seqIdx++) {
+      const seq = rawSeqs[seqIdx]!
+      const matches = matchChars[seqIdx]!
+      const insLens = insertLengths[seqIdx]!
+      const result: string[] = []
+
+      // Track position in original sequence for extracting inserts
+      let seqPos = 0
+
+      for (let pos = 0; pos < numPositions; pos++) {
+        const maxIns = maxInserts[pos]!
+
+        if (pos < insLens.length) {
+          const matchChar = matches[pos]
+          const insLen = insLens[pos]!
+
+          // Add match character
+          if (matchChar) {
+            result.push(matchChar)
+            seqPos++
+          } else {
+            result.push('.')
+          }
+
+          // Extract and uppercase inserts from original sequence
+          if (insLen > 0) {
+            result.push(seq.slice(seqPos, seqPos + insLen).toUpperCase())
+            seqPos += insLen
+          }
+
+          // Pad with gaps
+          const padding = maxIns - insLen
+          if (padding > 0) {
+            result.push(gapCache[padding]!)
+          }
+        } else {
+          // This sequence is shorter - add gaps
+          result.push(gapCache[1 + maxIns]!)
+        }
+      }
+
+      expanded[names[seqIdx]!] = result.join('')
+    }
+
+    return expanded
+  }
+
+  getMSA() {
+    return this.MSA
+  }
+
+  getRowData() {
+    return undefined
+  }
+
+  getNames() {
+    return this.orderedNames
+  }
+
+  getRow(name: string) {
+    return this.MSA.seqdata[name] || ''
+  }
+
+  getWidth() {
+    const name = Object.keys(this.MSA.seqdata)[0]
+    return name ? this.getRow(name).length : 0
+  }
+
+  getStructures() {
+    return {}
+  }
+
+  get alignmentNames() {
+    return []
+  }
+
+  getHeader() {
+    return {}
+  }
+
+  getTree(): NodeWithIds {
+    return {
+      id: 'root',
+      name: 'root',
+      noTree: true,
+      children: this.getNames().map(name => ({
+        id: name,
+        children: [],
+        name,
+      })),
+    }
+  }
+
+  get seqConsensus() {
+    return undefined
+  }
+
+  get secondaryStructureConsensus() {
+    return undefined
+  }
+
+  get tracks() {
+    return []
+  }
+}

package/src/parsers/ClustalMSA.ts

@@ -46,11 +46,13 @@ export default class ClustalMSA {
       id: 'root',
       name: 'root',
       noTree: true,
-      children: this.getNames().map(name => ({
-        id: name,
-        name,
-        children: [],
-      })),
+      children: this.getNames()
+        .filter((name): name is string => name !== undefined)
+        .map(name => ({
+          id: name,
+          name,
+          children: [],
+        })),
     }
   }
 

package/src/parsers/FastaMSA.ts

@@ -1,26 +1,26 @@
 import type { NodeWithIds } from '../types'
 
-function parseSmallFasta(text: string) {
-  return text
-    .split('>')
-    .filter(t => /\S/.test(t))
-    .map(entryText => {
-      const [defLine, ...seqLines] = entryText.split('\n')
-      const [id, ...description] = defLine!.split(' ')
-      const descriptionStr = description.join(' ')
-      const seqLinesStr = seqLines.join('')
-      const sequence = seqLinesStr.replaceAll(/\s/g, '')
-      return { id, description: descriptionStr, sequence }
-    })
-}
 export default class FastaMSA {
   private MSA: { seqdata: Record<string, string> }
+
   constructor(text: string) {
-    this.MSA = {
-      seqdata: Object.fromEntries(
-        parseSmallFasta(text).map(m => [m.id, m.sequence]),
-      ),
+    const seqdata: Record<string, string> = {}
+    for (const entry of text.split('>')) {
+      if (!/\S/.test(entry)) {
+        continue
+      }
+      const newlineIdx = entry.indexOf('\n')
+      if (newlineIdx === -1) {
+        continue
+      }
+      const defLine = entry.slice(0, newlineIdx)
+      const spaceIdx = defLine.indexOf(' ')
+      const id = spaceIdx === -1 ? defLine : defLine.slice(0, spaceIdx)
+      if (id) {
+        seqdata[id] = entry.slice(newlineIdx + 1).replaceAll(/\s/g, '')
+      }
     }
+    this.MSA = { seqdata }
   }
 
   getMSA() {