@gmod/bbi 7.1.0 → 8.0.1

package/src/index.ts

@@ -1,4 +1,11 @@
 import './bigint-polyfill/polyfill.ts'
 export { BigWig } from './bigwig.ts'
 export { BigBed } from './bigbed.ts'
-export type { BigWigHeader, Feature, RequestOptions } from './types.ts'
+export { parseBigWig } from './parse-bigwig.ts'
+export type {
+  BigWigFeatureArrays,
+  BigWigHeader,
+  Feature,
+  RequestOptions,
+  SummaryFeatureArrays,
+} from './types.ts'

package/src/parse-bigwig.ts

@@ -0,0 +1,19 @@
+import type { BigWig } from './bigwig.ts'
+import type { Feature, RequestOptions2 } from './types.ts'
+
+export async function parseBigWig(
+  bigwig: BigWig,
+  opts?: RequestOptions2,
+): Promise<Feature[]> {
+  const header = await bigwig.getHeader(opts)
+  const allFeatures: Feature[] = []
+
+  for (const ref of Object.values(header.refsByNumber)) {
+    const features = await bigwig.getFeatures(ref.name, 0, ref.length, opts)
+    for (const feature of features) {
+      allFeatures.push(feature)
+    }
+  }
+
+  return allFeatures
+}

package/src/range.ts

@@ -15,7 +15,7 @@ export default class Range {
   }
 
   get min() {
-    return this.ranges[0].min
+    return this.ranges[0]!.min
   }
 
   get max() {
@@ -36,7 +36,7 @@ export default class Range {
   }
 
   public getRanges() {
-    return this.ranges.map(r => new Range([{ min: r.min, max: r.max }]))
+    return this.ranges.map(r => new Range([r]))
   }
 
   public toString(): string {
@@ -44,32 +44,24 @@ export default class Range {
   }
 
   public union(s1: Range) {
-    const ranges = [...this.getRanges(), ...s1.getRanges()].sort((a, b) => {
-      if (a.min < b.min) {
-        return -1
-      } else if (a.min > b.min) {
-        return 1
-      } else if (a.max < b.max) {
-        return -1
-      } else if (b.max > a.max) {
-        return 1
-      } else {
-        return 0
-      }
+    const allRanges = [...this.ranges, ...s1.ranges].sort((a, b) => {
+      return a.min !== b.min ? a.min - b.min : a.max - b.max
     })
-    const oranges = [] as Range[]
-    let current = ranges[0]
 
-    for (const nxt of ranges) {
+    const merged: IRange[] = []
+    let current = allRanges[0]!
+
+    for (let i = 1; i < allRanges.length; i++) {
+      const nxt = allRanges[i]!
       if (nxt.min > current.max + 1) {
-        oranges.push(current)
+        merged.push(current)
         current = nxt
       } else if (nxt.max > current.max) {
-        current = new Range([{ min: current.min, max: nxt.max }])
+        current = { min: current.min, max: nxt.max }
       }
     }
-    oranges.push(current)
+    merged.push(current)
 
-    return oranges.length === 1 ? oranges[0] : new Range(oranges)
+    return new Range(merged)
   }
 }

package/src/types.ts

@@ -7,7 +7,7 @@ export interface ZoomLevel {
 
 export interface Feature {
   offset?: number
-  chromId: number
+  chromId?: number
   start: number
   end: number
   score?: number
@@ -46,7 +46,10 @@ export interface BigWigHeader {
   unzoomedDataOffset: number
   definedFieldCount: number
   uncompressBufSize: number
-  chromTreeOffset: number
+  // Offset to the B+ tree that maps chromosome names to integer IDs.
+  // This is part of the "cirTree" (combined ID R-tree) - a structure that
+  // combines a B+ tree for chromosome names with an R-tree for spatial queries
+  chromosomeTreeOffset: number
   extHeaderOffset: number
   fileType: string
 }
@@ -64,3 +67,17 @@ export interface RequestOptions2 extends RequestOptions {
   scale?: number
   basesPerSpan?: number
 }
+
+export interface BigWigFeatureArrays {
+  starts: Int32Array
+  ends: Int32Array
+  scores: Float32Array
+}
+
+export interface SummaryFeatureArrays {
+  starts: Int32Array
+  ends: Int32Array
+  scores: Float32Array
+  minScores: Float32Array
+  maxScores: Float32Array
+}

package/src/unzip.ts

@@ -1,5 +1,88 @@
-import { inflateRaw } from 'pako-esm2'
+import {
+  decompressAndParseBigWig,
+  decompressAndParseSummary,
+  inflateRawBatch,
+} from './wasm/inflate-wasm-inlined.js'
 
-export function unzip(input: Uint8Array) {
-  return inflateRaw(input.subarray(2), undefined)
+import type {
+  BigWigFeatureArrays,
+  SummaryFeatureArrays,
+} from './wasm/inflate-wasm-inlined.js'
+
+export interface UnzipBatchResult {
+  data: Uint8Array
+  offsets: number[]
+}
+
+export async function unzipBatch(
+  data: Uint8Array,
+  blocks: { offset: number; length: number }[],
+  maxOutputSize: number,
+): Promise<UnzipBatchResult> {
+  const inputOffsets = new Uint32Array(blocks.length)
+  const inputLengths = new Uint32Array(blocks.length)
+
+  for (let i = 0; i < blocks.length; i++) {
+    const block = blocks[i]!
+    inputOffsets[i] = block.offset
+    inputLengths[i] = block.length
+  }
+
+  return inflateRawBatch(data, inputOffsets, inputLengths, maxOutputSize)
+}
+
+export async function decompressAndParseBigWigBlocks(
+  data: Uint8Array,
+  blocks: { offset: number; length: number }[],
+  maxOutputSize: number,
+  reqStart: number,
+  reqEnd: number,
+): Promise<BigWigFeatureArrays> {
+  const inputOffsets = new Uint32Array(blocks.length)
+  const inputLengths = new Uint32Array(blocks.length)
+
+  for (let i = 0; i < blocks.length; i++) {
+    const block = blocks[i]!
+    inputOffsets[i] = block.offset
+    inputLengths[i] = block.length
+  }
+
+  return decompressAndParseBigWig(
+    data,
+    inputOffsets,
+    inputLengths,
+    maxOutputSize,
+    reqStart,
+    reqEnd,
+  )
 }
+
+export async function decompressAndParseSummaryBlocks(
+  data: Uint8Array,
+  blocks: { offset: number; length: number }[],
+  maxOutputSize: number,
+  reqChrId: number,
+  reqStart: number,
+  reqEnd: number,
+): Promise<SummaryFeatureArrays> {
+  const inputOffsets = new Uint32Array(blocks.length)
+  const inputLengths = new Uint32Array(blocks.length)
+
+  for (let i = 0; i < blocks.length; i++) {
+    const block = blocks[i]!
+    inputOffsets[i] = block.offset
+    inputLengths[i] = block.length
+  }
+
+  return decompressAndParseSummary(
+    data,
+    inputOffsets,
+    inputLengths,
+    maxOutputSize,
+    reqChrId,
+    reqStart,
+    reqEnd,
+  )
+}
+
+export type { BigWigFeatureArrays, SummaryFeatureArrays } from './types.ts'

package/src/util.ts

@@ -21,19 +21,18 @@ export function groupBlocks(blocks: Block[]) {
   let lastBlockEnd: number | undefined
   for (const block of blocks) {
     if (lastBlock && lastBlockEnd && block.offset - lastBlockEnd <= 2000) {
-      lastBlock.length =
-        lastBlock.length + block.length - lastBlockEnd + block.offset
+      lastBlock.length = block.offset + block.length - lastBlock.offset
       lastBlock.blocks.push(block)
+      lastBlockEnd = block.offset + block.length
     } else {
-      blockGroups.push(
-        (lastBlock = {
-          blocks: [block],
-          length: block.length,
-          offset: block.offset,
-        }),
-      )
+      lastBlock = {
+        blocks: [block],
+        length: block.length,
+        offset: block.offset,
+      }
+      blockGroups.push(lastBlock)
+      lastBlockEnd = block.offset + block.length
     }
-    lastBlockEnd = lastBlock.offset + lastBlock.length
   }
 
   return blockGroups

package/src/wasm/inflate-wasm-inlined.d.ts

@@ -0,0 +1,49 @@
+export function inflateRaw(
+  input: Uint8Array,
+  outputSize: number,
+): Promise<Uint8Array>
+export function inflateRawUnknownSize(input: Uint8Array): Promise<Uint8Array>
+
+export interface BatchResult {
+  data: Uint8Array
+  offsets: number[]
+}
+export function inflateRawBatch(
+  inputs: Uint8Array,
+  inputOffsets: Uint32Array,
+  inputLengths: Uint32Array,
+  maxOutputSize: number,
+): Promise<BatchResult>
+
+export interface BigWigFeatureArrays {
+  starts: Int32Array
+  ends: Int32Array
+  scores: Float32Array
+}
+
+export interface SummaryFeatureArrays {
+  starts: Int32Array
+  ends: Int32Array
+  scores: Float32Array
+  minScores: Float32Array
+  maxScores: Float32Array
+}
+
+export function decompressAndParseBigWig(
+  inputs: Uint8Array,
+  inputOffsets: Uint32Array,
+  inputLengths: Uint32Array,
+  maxBlockSize: number,
+  reqStart: number,
+  reqEnd: number,
+): Promise<BigWigFeatureArrays>
+
+export function decompressAndParseSummary(
+  inputs: Uint8Array,
+  inputOffsets: Uint32Array,
+  inputLengths: Uint32Array,
+  maxBlockSize: number,
+  reqChrId: number,
+  reqStart: number,
+  reqEnd: number,
+): Promise<SummaryFeatureArrays>