@gmod/cram

Read CRAM files (indexed or unindexed) with pure JS, works in node or in the browser.

• Reads CRAM 3.x and 2.x (3.1 added in v1.6.0)
• Does not read CRAM 1.x
• Can use .crai indexes out of the box, for efficient sequence fetching, but also has an index API that would allow use with other index types
• Has preliminary support for bzip2 and lzma codecs. lzma requires the latest @gmod/cram version, and uses webassembly. If you find you are unable to compile it, you can try downgrading

Install

$ npm install --save @gmod/cram
# or
$ yarn add @gmod/cram

Usage

const { IndexedCramFile, CramFile, CraiIndex } = require('@gmod/cram')

// Use indexedfasta library for seqFetch, if using local file (see below)
const { IndexedFasta, BgzipIndexedFasta } = require('@gmod/indexedfasta')

// this uses local file paths for node.js for IndexedFasta, for usages using
// remote URLs see indexedfasta docs for filehandles and
// https://github.com/gmod/generic-filehandle
const t = new IndexedFasta({
  path: '/filesystem/yourfile.fa',
  faiPath: '/filesystem/yourfile.fa.fai',
})

// example of fetching records from an indexed CRAM file.
// NOTE: only numeric IDs for the reference sequence are accepted.
// For indexedfasta the numeric ID is the order in which the sequence names
// appear in the header

// Wrap in an async and then run
run = async () => {
  const idToName = []
  const nameToId = {}

  // example opening local files on node.js
  // can also pass `cramUrl` (for the IndexedCramFile class), and `url` (for
  // the CraiIndex) params to open remote URLs
  //
  // alternatively `cramFilehandle` (for the IndexedCramFile class) and
  // `filehandle` (for the CraiIndex) can be used,  see for examples
  // https://github.com/gmod/generic-filehandle

  const indexedFile = new IndexedCramFile({
    cramPath: '/filesystem/yourfile.cram',
    //or
    //cramUrl: 'url/to/file.cram'
    //cramFilehandle: a generic-filehandle or similar filehandle
    index: new CraiIndex({
      path: '/filesystem/yourfile.cram.crai',
      // or
      // url: 'url/to/file.cram.crai'
      // filehandle: a generic-filehandle or similar filehandle
    }),
    seqFetch: async (seqId, start, end) => {
      // note:
      // * seqFetch should return a promise for a string, in this instance retrieved from IndexedFasta
      // * we use start-1 because cram-js uses 1-based but IndexedFasta uses 0-based coordinates
      // * the seqId is a numeric identifier, so we convert it back to a name with idToName
      // * you can return an empty string from this function for testing if you want, but you may not get proper interpretation of record.readFeatures
      return t.getSequence(idToName[seqId], start - 1, end)
    },
    checkSequenceMD5: false,
  })
  const samHeader = await indexedFile.cram.getSamHeader()

  // use the @SQ lines in the header to figure out the
  // mapping between ref ref ID numbers and names

  const sqLines = samHeader.filter(l => l.tag === 'SQ')
  sqLines.forEach((sqLine, refId) => {
    sqLine.data.forEach(item => {
      if (item.tag === 'SN') {
        // this is the ref name
        const refName = item.value
        nameToId[refName] = refId
        idToName[refId] = refName
      }
    })
  })

  const records = await indexedFile.getRecordsForRange(
    nameToId['chr1'],
    10000,
    20000,
  )
  records.forEach(record => {
    console.log(`got a record named ${record.readName}`)
    if (record.readFeatures != undefined) {
      record.readFeatures.forEach(({ code, pos, refPos, ref, sub }) => {
        // process the read features. this can be used similar to
        // CIGAR/MD strings in SAM. see CRAM specs for more details.
        if (code === 'X') {
          console.log(
            `${record.readName} shows a base substitution of ${ref}->${sub} at ${refPos}`,
          )
        }
      })
    }
  })
}

run()

// can also pass `cramUrl` (for the IndexedCramFile class), and `url` (for the CraiIndex) params to open remote URLs
// alternatively `cramFilehandle` (for the IndexedCramFile class) and `filehandle` (for the CraiIndex) can be used,  see for examples https://github.com/gmod/generic-filehandle

You can use cram-js without NPM also with the cram-bundle.js. See the example directory for usage with script tag

API (auto-generated)

• CramRecord - format of CRAM records returned by this API
  • ReadFeatures - format of read features on records
• IndexedCramFile - indexed access into a CRAM file
• CramFile - .cram API
• CraiIndex - .crai index API
• Error Classes - special error classes thrown by this API

CramRecord

Table of Contents

• CramRecord
  • isPaired
  • isProperlyPaired
  • isSegmentUnmapped
  • isMateUnmapped
  • isReverseComplemented
  • isMateReverseComplemented
  • isRead1
  • isRead2
  • isSecondary
  • isFailedQc
  • isDuplicate
  • isSupplementary
  • isDetached
  • hasMateDownStream
  • isPreservingQualityScores
  • isUnknownBases
  • getReadBases
  • getPairOrientation
  • addReferenceSequence
    • Parameters

CramRecord

Class of each CRAM record returned by this API.

isPaired

Returns boolean true if the read is paired, regardless of whether both segments are mapped

isProperlyPaired

Returns boolean true if the read is paired, and both segments are mapped

isSegmentUnmapped

Returns boolean true if the read itself is unmapped; conflictive with isProperlyPaired

isMateUnmapped

Returns boolean true if the read itself is unmapped; conflictive with isProperlyPaired

isReverseComplemented

Returns boolean true if the read is mapped to the reverse strand

isMateReverseComplemented

Returns boolean true if the mate is mapped to the reverse strand

isRead1

Returns boolean true if this is read number 1 in a pair

isRead2

Returns boolean true if this is read number 2 in a pair

isSecondary

Returns boolean true if this is a secondary alignment

isFailedQc

Returns boolean true if this read has failed QC checks

isDuplicate

Returns boolean true if the read is an optical or PCR duplicate

isSupplementary

Returns boolean true if this is a supplementary alignment

isDetached

Returns boolean true if the read is detached

hasMateDownStream

Returns boolean true if the read has a mate in this same CRAM segment

isPreservingQualityScores

Returns boolean true if the read contains qual scores

isUnknownBases

Returns boolean true if the read has no sequence bases

getReadBases

Get the original sequence of this read.

Returns String sequence basepairs

getPairOrientation

Get the pair orientation of a paired read. Adapted from igv.js

Returns String of paired orientatin

addReferenceSequence

Annotates this feature with the given reference sequence basepair information. This will add a sub and a ref item to base subsitution read features given the actual substituted and reference base pairs, and will make the getReadSequence() method work.

Parameters

• refRegion object

  • refRegion.start number
  • refRegion.end number
  • refRegion.seq string

• compressionScheme CramContainerCompressionScheme

Returns undefined nothing

ReadFeatures

The feature objects appearing in the readFeatures member of CramRecord objects that show insertions, deletions, substitutions, etc.

Static fields

• code (character): One of "bqBXIDiQNSPH". See page 15 of the CRAM v3 spec for their meanings.
• data (any): the data associated with the feature. The format of this varies depending on the feature code.
• pos (number): location relative to the read (1-based)
• refPos (number): location relative to the reference (1-based)

IndexedCramFile

Table of Contents

• constructor
  • Parameters
• getRecordsForRange
  • Parameters
• hasDataForReferenceSequence
  • Parameters

constructor

Parameters

• args object

  • args.cram CramFile
  • args.index Index-like object that supports getEntriesForRange(seqId,start,end) -> Promise[Array[index entries]]
  • args.cacheSize number? optional maximum number of CRAM records to cache. default 20,000
  • args.checkSequenceMD5 boolean? default true. if false, disables verifying the MD5 checksum of the reference sequence underlying a slice. In some applications, this check can cause an inconvenient amount (many megabases) of sequences to be fetched.

getRecordsForRange

Parameters

• seq number numeric ID of the reference sequence
• start number start of the range of interest. 1-based closed coordinates.
• end number end of the range of interest. 1-based closed coordinates.
• opts (optional, default {})

hasDataForReferenceSequence

Parameters

• seqId number

Returns Promise true if the CRAM file contains data for the given reference sequence numerical ID

CramFile

Table of Contents

• constructor
  • Parameters
• containerCount

constructor

Parameters

• args object

  • args.filehandle object? a filehandle that implements the stat() and read() methods of the Node filehandle API https://nodejs.org/api/fs.html#fs_class_filehandle
  • args.path object? path to the cram file
  • args.url object? url for the cram file. also supports file:// urls for local files
  • args.seqFetch function? a function with signature (seqId, startCoordinate, endCoordinate) that returns a promise for a string of sequence bases
  • args.cacheSize number? optional maximum number of CRAM records to cache. default 20,000
  • args.checkSequenceMD5 boolean? default true. if false, disables verifying the MD5 checksum of the reference sequence underlying a slice. In some applications, this check can cause an inconvenient amount (many megabases) of sequences to be fetched.

containerCount

CraiIndex

Table of Contents

• constructor
  • Parameters
• hasDataForReferenceSequence
  • Parameters
• getEntriesForRange
  • Parameters

constructor

Parameters

• args object

  • args.path string?
  • args.url string?
  • args.filehandle FileHandle?

hasDataForReferenceSequence

Parameters

• seqId number

Returns Promise true if the index contains entries for the given reference sequence ID, false otherwise

getEntriesForRange

fetch index entries for the given range

Parameters

• seqId number
• queryStart number
• queryEnd number

Returns Promise promise for an array of objects of the form {start, span, containerStart, sliceStart, sliceBytes }

CramUnimplementedError

Extends Error

Error caused by encountering a part of the CRAM spec that has not yet been implemented

CramMalformedError

Extends CramError

An error caused by malformed data.

CramBufferOverrunError

Extends CramMalformedError

An error caused by attempting to read beyond the end of the defined data.

Academic Use

This package was written with funding from the NHGRI as part of the JBrowse project. If you use it in an academic project that you publish, please cite the most recent JBrowse paper, which will be linked from jbrowse.org.

License

MIT © Robert Buels