bionode-ncbi - npm Package Compare versions

Comparing version 0.2.1 to 0.3.0

lib/bionode-ncbi.js

@@ -26,2 +26,3 @@ // # bionode-ncbi
 var fs = require('fs')
+var path = require('path')
 var mkdirp = require('mkdirp')
@@ -38,16 +39,22 @@ var async = require('async')
 var pumpify = require('pumpify')
+var URL = require('url')
+var cheerio = require('cheerio')
 
+var ncbi = exports
 
-module.exports = exports = ncbi = new NCBI()
-
-function NCBI() {
-  this.APIROOT = 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/'
-  this.DEFAULTS = 'retmode=json&version=2.0'
-  this.RETURNMAX = 50
-  this.XMLPROPERTIES = {
-    'sra': ['expxml', 'runs'],
-    'biosample': ['sampledata'],
-    'assembly': ['meta' ]
+var APIROOT = 'http://eutils.ncbi.nlm.nih.gov/entrez/eutils/'
+var DEFAULTS = 'retmode=json&version=2.0'
+var RETURNMAX = 50
+var XMLPROPERTIES = {
+  'sra': ['expxml', 'runs'],
+  'biosample': ['sampledata'],
+  'assembly': ['meta' ]
+}
+var LASTSTREAM = {
+  'sra': function() {
+    return pumpify.obj(
+      tool.ensureIsArray('runs.Run'),
+      tool.filterObjectsArray('total_bases', '', 'runs.Run')
+    )
   }
-  return this
 }
@@ -87,21 +94,13 @@
 
-NCBI.prototype.search = function(db, term, cb) {
-  var xmlProperties = ncbi.XMLPROPERTIES[db] || through.obj()
+ncbi.search = function(db, term, cb) {
+  return function() {
+  var xmlProperties = XMLPROPERTIES[db] || through.obj()
+  var lastStream = LASTSTREAM[db] || through.obj
 
-  var finish
-  if (db === 'sra') {
-    finish = pumpify.obj(
-      tool.ensureIsArray('runs.Run'),
-      tool.filterObjectsArray('total_bases', '', 'runs.Run')
-    )
-  }
-  else {
-    finish = through.obj()
-  }
-
   var stream = pumpify.obj(
-    _searchURL(db, term),
-    _requestStream(),
-    _dataURL(),
-    _requestStream(),
+    createAPISearchUrl(db, term),
+    requestStream(),
+    createAPIDataURL(),
+    requestStream(),
+    filterEmptyResults(),
     tool.extractProperty('result'),
@@ -111,3 +110,3 @@ tool.deleteProperty('uids'),
     tool.XMLToJSProperties(xmlProperties),
-    finish
+    lastStream()
   )
@@ -118,10 +117,14 @@
   else { return stream }
+  }()
 }
 
-function _searchURL(db, term) {
+
+function createAPISearchUrl(db, term) {
   var stream = through.obj(transform)
+  return stream
+
   function transform(obj, enc, next) {
     var query = [
-      ncbi.APIROOT + 'esearch.fcgi?',
-      ncbi.DEFAULTS,
+      APIROOT + 'esearch.fcgi?',
+      DEFAULTS,
       'db=' + db,
@@ -135,20 +138,20 @@ 'term=' + encodeURI(obj),
   }
-  return stream
 }
 
-function _dataURL() {
+function createAPIDataURL() {
   var stream = through.obj(transform)
   return stream
+
   function transform(obj, enc, next) {
     var count = obj.esearchresult.count
-    var numRequests = Math.floor(count / ncbi.RETURNMAX)
+    var numRequests = Math.floor(count / RETURNMAX)
     for (var i = 0; i <= numRequests; i++) {
-      var retstart = i * ncbi.RETURNMAX
+      var retstart = i * RETURNMAX
       var query = [
-        ncbi.APIROOT + 'esummary.fcgi?',
-        ncbi.DEFAULTS,
+        APIROOT + 'esummary.fcgi?',
+        DEFAULTS,
         'db=' + obj.db,
         'query_key=1',
         'WebEnv=' + obj.esearchresult.webenv,
-        'retmax=' + ncbi.RETURNMAX,
+        'retmax=' + RETURNMAX,
         'retstart=' + retstart
@@ -163,19 +166,18 @@ ].join('&')
 
-function _requestStream() {
+function filterEmptyResults() {
   var stream = through.obj(transform)
-  var outStream = through.obj()
+  return stream
+
   function transform(obj, enc, next) {
-    var self = this
-    request({ uri: obj, json: true }, gotData)
-    function gotData(err, res, body) {
-      debug('request response', res.statusCode)
-      debug('request results', body)
-      if (body.esearchresult && body.esearchresult.ERROR) {
-        self.emit('error', new Error(body.esearchresult.ERROR))
-      }
-      self.push(body)
-      next()
+    if (obj.esummaryresult && obj.esummaryresult[0] === 'Empty result - nothing todo') {
+      return next()
     }
+    if (obj.error && obj.error[0] === 'Empty result - nothing todo') {
+      return next()
+    }
+    if (obj.result) {
+      this.push(obj)
+    }
+    next()
   }
-  return stream
 }
@@ -201,41 +203,55 @@
 
-NCBI.prototype.link = function(srcDB, destDB, srcUID, cb) {
-  var stream = through.obj(getDestUID)
+ncbi.link = function(srcDB, destDB, srcUID, cb) {
+  var stream = pumpify.obj(
+    createAPILinkURL(srcDB, destDB),
+    requestStream(true),
+    createLinkObj()
+  )
+
   if (srcUID) { stream.write(srcUID); stream.end() }
   if (cb) { stream.pipe(concat(cb)) }
   else { return stream }
+}
 
-  function getDestUID(srcUID, enc, next) {
-    var self = this
+function createAPILinkURL(srcDB, destDB) {
+  var stream = through.obj(transform)
+  return stream
 
+  function transform(obj, enc, next) {
     var query = [
-      ncbi.APIROOT + 'elink.fcgi?',
+      APIROOT + 'elink.fcgi?',
       'dbfrom=' + srcDB,
       'db=' + destDB,
-      'id=' + srcUID
+      'id=' + obj.toString()
     ].join('&')
+    this.push(query)
+    next()
+  }
+}
 
-    var link = {
-      srcDB: srcDB,
-      destDB: destDB,
-      srcUID: srcUID
+function createLinkObj() {
+  var stream = through.obj(transform)
+  return stream
+
+  function transform(obj, enc, next) {
+    var self = this
+    var query = URL.parse(obj.url, true).query
+    var result = {
+      srcDB: query.dbfrom,
+      destDB: query.db,
+      srcUID: query.id
     }
-
-    var linkName = srcDB+'_'+destDB
-
-    var getLink = tool.attachToObject(link, 'destUID')
-
-    debug('elink request', query)
-
-    request({ uri: query, json: true })
-    .pipe(_wait(500))
-    .pipe(tool.XMLToJS(true))
-    .pipe(tool.extractProperty('LinkSet.0.LinkSetDb'))
-    .pipe(tool.arraySplit())
-    .pipe(tool.collectMatch('LinkName.0', linkName))
-    .pipe(tool.extractProperty('Link'))
-    .pipe(tool.arraySplit())
-    .pipe(tool.extractProperty('Id.0'))
-    .pipe(getLink)
-    _attachStandardEvents(getLink, self, next)
+    xml2js(obj.body, function(err, data) {
+      if (err) { self.emit('error', err) }
+      if (data.eLinkResult.LinkSet[0].LinkSetDb) {
+        data.eLinkResult.LinkSet[0].LinkSetDb.forEach(getMatch)
+        function getMatch(link) {
+          if (link.LinkName[0] === query.dbfrom + '_' + query.db) {
+            result.destUID = link.Link[0].Id[0]
+            self.push(result)
+          }
+        }
+      }
+      next()
+    })
   }
@@ -260,17 +276,66 @@ }
 
-NCBI.prototype.download = function(db, term, cb) {
-  var stream = through.obj(transform)
+ncbi.download = function(db, term, cb) {
+  var stream = pumpify.obj(
+    ncbi.urls(db),
+    download(db)
+  )
   if (term) { stream.write(term); stream.end() }
   if (cb) { stream.pipe(concat(cb)) }
   else { return stream }
+}
+
+function download(db, term) {
+  var stream = through.obj(transform)
+  return stream
+
   function transform(obj, enc, next) {
     var self = this
-    var download = _download()
-    var searchdb = db === 'gff' ? 'genome' : db
-    var getdb = db
+    var prevTime = Date.now()
+    var currTime
+    var chunkSizeMB = 1
+    var chunkSize = chunkSizeMB * 1024 * 1024 //bytes
+    var folder = obj.uid + '/'
+    var path = folder + obj.genomic.fna.replace(/.*\//, '')
 
-    ncbi.search(searchdb, obj)
-    .pipe(_getURLs(getdb))
-    .pipe(download)
-    _attachStandardEvents(download, self, next)
+    var log = {
+      url: obj.genomic.fna,
+      path: path,
+    }
+
+    mkdirp.sync(obj.uid)
+    if (!fs.existsSync(path)) {
+      debug('downloading', obj.url)
+
+
+      dld(obj.genomic.fna, folder, chunkSize)
+      .on('data', logging)
+      .on('end', function() {
+        log.status = 'completed'
+        log.speed = 'NA'
+        self.push(log)
+        next()
+      })
+      .on('error', function(err) { self.emit('error', err) })
+    }
+    else {
+      log.status = 'completed'
+      log.speed = 'NA'
+      log.size = Math.round(fs.statSync(path).size / 1024 / 1024) + ' MB'
+      self.push(log)
+      next()
+    }
+
+    function logging(position, size) {
+      var progress = (position * 100 / size).toFixed(2) + ' %'
+      var sizeMB = Math.round(size / 1024 / 1024) + ' MB'
+      currTime = Date.now()
+      var diffTimeSec = (currTime - prevTime) / 1000
+      prevTime = currTime
+      var speed =  (chunkSizeMB / diffTimeSec).toFixed(2) + ' MB/s'
+      log.status = 'downloading'
+      log.total = sizeMB
+      log.progress = progress
+      log.speed = speed
+      self.push(log)
+    }
   }
@@ -291,20 +356,17 @@ }
 
-NCBI.prototype.urls = function(db, term, cb) {
-  var stream = through.obj(transform)
+ncbi.urls = function(db, term, cb) {
+  var searchdb = db === 'gff' ? 'genome' : db
+  var stream = pumpify.obj(
+    ncbi.search(searchdb),
+    createFTPURL(db)
+  )
   if (term) { stream.write(term); stream.end() }
   if (cb) { stream.pipe(concat(cb)) }
   else { return stream }
-  function transform(obj, enc, next) {
-    var self = this
-    var searchdb = db === 'gff' ? 'genome' : db
-    var getdb = db
-    var getURLs = _getURLs(getdb)
-    ncbi.search(searchdb, obj)
-    .pipe(getURLs)
-    _attachStandardEvents(getURLs, self, next)
-  }
 }
 
-function _getURLs(db) {
-  return through.obj(transform)
+function createFTPURL(db) {
+  var stream = through.obj(transform)
+  return stream
+
   function transform(obj, enc, next) {
@@ -341,8 +403,27 @@ var self = this
         var ftpArray = Array.isArray(ftpPath) ? ftpPath : [ ftpPath ]
-        ftpArray.forEach(function(ftp) {
-          var assemblyURL = ftp._.replace('ftp://', 'http://') + 'Primary_Assembly/unplaced_scaffolds/FASTA/unplaced.scaf.fa.gz'
-          self.push({url: assemblyURL, uid: obj.uid})
-        })
+        var httpRoot = ftpArray[0]._.replace('ftp://', 'http://') // NCBI seems to return GenBank and RefSeq accessions for the same thing. We only need one.
+        request(httpRoot, gotFTPDir)
+        function gotFTPDir(err, res, body) {
+          if (err) { self.emit('error', err) }
+          if (!res || res.statusCode !== 200) { self.emit('err', res) }
+          if (!body) { return next() }
+          $ = cheerio.load(body)
+
+          var urls = { uid: obj.uid }
+
+          $('a').map(attachToResult)
+          function attachToResult(i, a) {
+            var href = a.attribs.href
+            var base = path.basename(href)
+            var fileNameProperties = base.replace(/.*\//, '').split('_')
+            var fileNameExtensions = fileNameProperties[fileNameProperties.length-1].split('.')
+            var fileType = fileNameExtensions[0]
+            var fileFormat = fileNameExtensions[1] || 'dir'
+            if (!urls[fileType]) { urls[fileType] = {} }
+            urls[fileType][fileFormat] = httpRoot + '/' + href
+          }
+          self.push(urls)
+          next()
+        }
       }
-      next()
     }
@@ -368,60 +449,27 @@
 
-function _download(db, term) {
-  return through.obj(transform)
-  function transform(obj, enc, next) {
-    var self = this
-    var prevTime = Date.now()
-    var currTime
-    var chunkSizeMB = 1
-    var chunkSize = chunkSizeMB * 1024 * 1024 //bytes
-    var folder = obj.uid + '/'
-    var path = folder + obj.url.replace(/.*\//, '')
 
-    mkdirp.sync(obj.uid)
-    if (!fs.existsSync(path)) {
-      debug('downloading', obj.url)
-      dld(obj.url, folder, chunkSize)
-      .on('data', log)
-      .on('end', function() {
-        self.push(path)
-        next()
-      })
-      .on('error', function(err) { self.emit('error', err) })
-    }
-    else {
-      self.push(path)
-      next()
-    }
+function requestStream(returnURL) {
+  var stream = through.obj(transform)
+  return stream
 
-    function log(position, size) {
-      var progress = (position * 100 / size).toFixed(2) + ' %'
-      var sizeMB = Math.round(size / 1024 / 1024) + ' MB'
-      currTime = Date.now()
-      var diffTimeSec = (currTime - prevTime) / 1000
-      prevTime = currTime
-      var speed =  (chunkSizeMB / diffTimeSec).toFixed(2) + ' MB/s'
-      var log = 'Downloading ' + path+' '+ progress + ' of ' + sizeMB + ' at ' + speed
-      self.push(log)
-    }
-  }
-}
-
-
-function _attachStandardEvents(stream, self, next) {
-  stream
-  .on('data', function(data) { self.push(data) })
-  .on('end', function() { next() })
-  .on('error', function(err) { self.emit('error', err) })
-}
-
-function _wait(ms) {
-  return through.obj(transform)
   function transform(obj, enc, next) {
     var self = this
-    setTimeout(pushObj, ms)
-    function pushObj() {
-      self.push(obj)
-      next()
+    get()
+    self.tries = 1
+    function get() {
+      if (self.tries > 20) { console.warn('tries' + self.tries + obj) }
+      request({ uri: obj, json: true, timeout: 5000 }, gotData)
+      function gotData(err, res, body) {
+        if (err || !res) { self.tries++; return get() }
+        debug('request response', res.statusCode)
+        debug('request results', body)
+        if (body.esearchresult && body.esearchresult.ERROR) {
+          self.emit('error', new Error(body.esearchresult.ERROR))
+        }
+        var result = returnURL ? {url: obj, body: body} : body
+        self.push(result)
+        next()
+      }
     }
   }
 }

package.json

 {
   "name": "bionode-ncbi",
   "description": "Node.js module for working with the NCBI API (aka e-utils) using Streams.",
-  "version": "0.2.1",
+  "version": "0.3.0",
   "homepage": "http://github.com/bionode/bionode-ncbi",
@@ -17,2 +17,5 @@ "repository": {
     "async": "^0.9.0",
+    "cheerio": "^0.17.0",
+    "concat-stream": "~1.4.6",
+    "debug": "^1.0.3",
     "dld": "0.0.2",
@@ -22,11 +25,10 @@ "docco": "^0.6.3",
     "nested-property": "0.0.2",
+    "pumpify": "^1.3.2",
     "request": "^2.37.0",
     "through2": "^0.5.1",
     "tool-stream": "0.0.1",
-    "debug": "^1.0.3",
-    "xml2js": "^0.4.4",
-    "concat-stream": "~1.4.6",
-    "pumpify": "^1.3.2"
+    "xml2js": "^0.4.4"
   },
   "devDependencies": {
+    "contributor": "~0.1.16",
     "coveralls": "~2.11.1",
@@ -38,3 +40,5 @@ "docco": "~0.6.3",
     "should": "~4.0.4",
-    "contributor": "~0.1.16"
+    "tap-spec": "^0.2.1",
+    "tape": "^2.14.0",
+    "testling": "^1.7.0"
   },
@@ -54,5 +58,5 @@ "keywords": [
   "scripts": {
-    "test": "mocha --reporter spec && rm -rf ./503988",
-    "coverage": "istanbul cover ./node_modules/.bin/_mocha --report lcovonly -- -R spec && rm -rf ./coverage",
-    "coveralls": "istanbul cover ./node_modules/.bin/_mocha --report lcovonly -- -R spec && cat ./coverage/lcov.info | ./node_modules/.bin/coveralls && rm -rf ./coverage",
+    "test": "node test/*.js | tap-spec && rm -rf ./503988",
+    "coverage": "istanbul cover test/bionode-ncbi.js --report lcovonly -- | tap-spec && rm -rf ./coverage",
+    "coveralls": "istanbul cover test/bionode-ncbi.js --report lcovonly -- | tap-spec && cat ./coverage/lcov.info | ./node_modules/.bin/coveralls && rm -rf ./coverage",
     "build-docs": "docco ./lib/bionode-ncbi.js"
@@ -59,0 +63,0 @@ },

test/bionode-ncbi.js

 var fs = require('fs')
 var crypto = require('crypto')
+var async = require('async')
+var test = require('tape')
 
 var ncbi = require('../')
+
 var testData = require('./data')
 var guillardiaThetaSRAData = require('./guillardia-theta.sra')
-var should = require('should')
 
-require('mocha')
 
+test('Download list', function(t) {
+  t.plan(2)
 
-describe("Download list", function() {
-  this.timeout(60000)
-  it("should take a database name (assembly) and search term (Guillardia theta), and list datasets URLs", function(done) {
-    ncbi.urls('assembly', 'Guillardia theta').on('data', function(data) {
-      data.should.eql(testData.assembly['guillardia-theta']['download-list'][0])
-      done()
+    ncbi.urls('assembly', 'Guillardia theta')
+    .on('data', function(data) {
+      var msg = 'should take a database name (assembly) and search term (Guillardia theta), and list datasets URLs'
+      t.deepEqual(data, testData.assembly['guillardia-theta'].urls, msg)
     })
-  })
-  it("should take a database name (sra) and search term (Guillardia theta), and list datasets URLs", function(done) {
+
     var results = []
     ncbi.urls('sra', 'Guillardia theta')
-    .on('data', function(data) {
-      results.push(data)
-    })
+    .on('data', function(data) { results.push(data) })
     .on('end', function(data) {
-      results.should.eql(testData.sra['guillardia-theta']['download-list'])
-      done()
+      var msg = 'should take a database name (sra) and search term (Guillardia theta), and list datasets URLs'
+      t.deepEqual(results, testData.sra['guillardia-theta'].urls, msg)
     })
-  })
 })
 
-describe("Download", function() {
-  this.timeout(600000)
-  it("should take a database name and search term, and download datasets", function(done) {
-    download(done)
-  })
-  it("repeat same download to cover already downloaded branch", function(done) {
-    download(done)
-  })
-  function download(cb) {
+
+test('Download', function(t) {
+  t.plan(2)
+
+  async.eachSeries(
+    [
+      'should take a database name and search term, and download datasets',
+      'repeat same download to cover already downloaded branch'
+    ],
+    testDownload
+  )
+
+  function testDownload(msg, cb) {
     var path
     ncbi.download('assembly', 'Guillardia theta')
-    .on('data', function(data) { path = data })
+    .on('data', function(data) { path = data.path })
     .on('end', function() {
@@ -51,3 +52,3 @@ var file = fs.ReadStream(path)
         var sha1 = shasum.digest('hex');
-        sha1.should.eql('3f9fc7e9698596ca6ba8eb85fed1f2d7a1070995')
+        t.equal(sha1, 'a2dc7b3b0ae6f40d5205c4394c2fe8bc65d52bc2', msg)
         cb()
@@ -59,45 +60,39 @@ })
 
-describe("Search", function() {
-  this.timeout(60000)
-  it("should take a database name and search term, and return the data", function(done) {
-    ncbi.search('assembly', 'Guillardia theta')
-    .on('data', function (data) {
-      data.should.eql(testData.assembly['guillardia-theta'].search)
-      done()
-    })
+
+test('Search', function(t) {
+  t.plan(2)
+
+  ncbi.search('assembly', 'Guillardia theta')
+  .on('data', function (data) {
+    var msg = 'should take a database name and search term, and return the data'
+    t.deepEqual(data, testData.assembly['guillardia-theta'].search, msg)
   })
-  it("same as previous but searching sra instead of assembly", function(done) {
-    var results = []
-    ncbi.search('sra', 'Guillardia theta')
-    .on('data', function(data) {
-      results.push(JSON.parse(JSON.stringify(data)))
-    })
-    .on('end', function() {
-      results.should.eql(guillardiaThetaSRAData)
-      done()
-    })
+
+  var results = []
+  ncbi.search('sra', 'Guillardia theta')
+  .on('data', function(data) { results.push(data) })
+  .on('end', function() {
+    var msg = 'same as previous but searching sra instead of assembly'
+    t.deepEqual(results, guillardiaThetaSRAData, msg)
   })
 })
 
-describe("Link", function() {
-  this.timeout(60000)
-  it("should take names for source database, destination database and a NCBI UID, and return the link", function(done) {
-    var results = []
-    ncbi.link('sra', 'bioproject', '35533')
-    .on('data', function(data) {
-      results.push(JSON.parse(JSON.stringify(data)))
-    })
-    .on('end', function() {
-      results.should.eql(testData.link['sra-bioproject']['35533'])
-      done()
-    })
+
+test('Link', function(t) {
+  t.plan(2)
+
+  var results = []
+  ncbi.link('sra', 'bioproject', '35533')
+  .on('data', function(data) { results.push(data) })
+  .on('end', function() {
+    var msg = 'should take names for source database, destination database and a NCBI UID, and return the link'
+    t.deepEqual(results, testData.link['sra-bioproject']['35533'], msg)
   })
-  it("same as previous, but doing bioproject->assembly instead of sra->assembly to try get same assembly UID as Search", function(done) {
+
+  ncbi.link('bioproject', 'assembly', '53577')
+  .on('data', function(data) {
     var results = []
-    ncbi.link('bioproject', 'assembly', '53577')
-    .on('data', function(data) {
-      data.destUID.should.eql(testData.assembly['guillardia-theta'].search.uid)
-      done()
-    })
+    var msg = 'same as previous, but doing bioproject->assembly instead of sra->assembly to try get same assembly UID as Search'
+    t.deepEqual(data.destUID, testData.assembly['guillardia-theta'].search.uid, msg)
   })
 })

test/data.json

 {
   "assembly": {
     "guillardia-theta": {
-      "search": {"uid":"503988","rsuid":"1011608","gbuid":"503988","assemblyaccession":"GCF_000315625.1","chainid":"315625","assemblyname":"Guith1","ucscname":"","ensemblname":"","assemblydescription":"","taxid":"905079","organism":"Guillardia theta CCMP2712","speciestaxid":"55529","speciesname":"Guillardia theta","assemblyclass":"haploid","wgs":"AEIE01","releaselevel":"Major","rs_projects":["223305"],"gb_projects":["53577"],"rs_bioprojects":[{"bioprojectid":223305,"bioprojectaccn":"PRJNA223305"}],"gb_bioprojects":[{"bioprojectid":53577,"bioprojectaccn":"PRJNA53577"}],"biosampleaccn":"SAMN00116900","biosampleid":"116900","biosource":{"isolate":"","sex":"","infraspecieslist":[{"sub_type":"strain","sub_value":"CCMP2712"}]},"coverage":"0","partialgenomerepresentation":"false","asmreleasedate":"2012/12/06 00:00","seqreleasedate":"2012/12/05 00:00","asmupdatedate":"2014/04/22 00:00","ncbireleasedate":"2012/12/06 00:00","submissiondate":"2012/12/05 00:00","lastmajorreleaseaccession": "GCF_000315625.1","lastupdatedate":"2014/04/22 00:00","primary":"1011598","propertylist":["full-genome-representation","latest","latest_genbank","latest_refseq","representative","species-default","wgs"],"synonym":{"genbank":"GCA_000315625.1","refseq":"GCF_000315625.1","similarity":"identical"},"sortorder":"2C50003156259898","submitterorganization":"JGI","refseq_category":"representative-genome","anomalouslist":[],"assemblystatus":"Scaffold","meta":{"Stats":{"Stat":[{"_":"0","category":"alt_loci_count","sequence_tag":"all"},{"_":"0","category":"chromosome_count","sequence_tag":"all"},{"_":"5126","category":"contig_count","sequence_tag":"all"},{"_":"587","category":"contig_l50","sequence_tag":"all"},{"_":"40445","category":"contig_n50","sequence_tag":"all"},{"_":"0","category":"non_chromosome_replicon_count","sequence_tag":"all"},{"_":"0","category":"replicon_count","sequence_tag":"all"},{"_":"669","category":"scaffold_count","sequence_tag":"all"},{"_":"0","category":"scaffold_count","sequence_tag":"placed"},{"_":"0","category":"scaffold_count","sequence_tag":"unlocalized"},{"_":"669","category":"scaffold_count","sequence_tag":"unplaced"},{"_":"52","category":"scaffold_l50","sequence_tag":"all"},{"_":"545808","category":"scaffold_n50","sequence_tag":"all"},{"_":"87145349","category":"total_length","sequence_tag":"all"},{"_":"83457412","category":"ungapped_length","sequence_tag":"all"}]},"FtpSites":{"FtpPath":{"_":"ftp://ftp.ncbi.nlm.nih.gov/genbank/genomes/Eukaryotes/protozoa/Guillardia_theta/Guith1/","type":"GenBank"}},"assembly-level":"2","assembly-status":"Scaffold","representative-status":"representative-genome","submitter-organization":"JGI"}},
-      "download-list": [
-        {"url": "http://ftp.ncbi.nlm.nih.gov/genbank/genomes/Eukaryotes/protozoa/Guillardia_theta/Guith1/Primary_Assembly/unplaced_scaffolds/FASTA/unplaced.scaf.fa.gz", "uid": "503988"}
-      ]
+      "search": {
+        "uid":"503988",
+        "rsuid":"1011608",
+        "gbuid":"503988",
+        "assemblyaccession":"GCF_000315625.1",
+        "lastmajorreleaseaccession":"GCF_000315625.1",
+        "chainid":"315625",
+        "assemblyname":"Guith1",
+        "ucscname":"",
+        "ensemblname":"",
+        "taxid":"905079",
+        "organism":"Guillardia theta CCMP2712",
+        "speciestaxid":"55529",
+        "speciesname":"Guillardia theta",
+        "assemblyclass":"haploid",
+        "assemblystatus":"Scaffold",
+        "wgs":"AEIE01",
+        "gb_bioprojects":[
+          {
+            "bioprojectaccn":"PRJNA53577",
+            "bioprojectid":53577
+          }
+        ],
+        "gb_projects":[
+          "53577"
+        ],
+        "rs_bioprojects":[
+          {
+           "bioprojectaccn":"PRJNA223305",
+           "bioprojectid":223305
+          }
+        ],
+        "rs_projects":[
+          "223305"
+        ],
+        "biosampleaccn":"SAMN00116900",
+        "biosampleid":"116900",
+        "biosource":{
+          "infraspecieslist":[
+            {
+              "sub_type":"strain",
+              "sub_value":"CCMP2712"
+            }
+          ],
+          "sex":"",
+          "isolate":""
+        },
+        "coverage":"0",
+        "partialgenomerepresentation":"false",
+        "primary":"1011598",
+        "assemblydescription":"",
+        "releaselevel":"Major",
+        "asmreleasedate":"2012/12/06 00:00",
+        "seqreleasedate":"2012/12/05 00:00",
+        "asmupdatedate":"2014/04/22 00:00",
+        "ncbireleasedate":"2012/12/06 00:00",
+        "submissiondate":"2012/12/05 00:00",
+        "lastupdatedate":"2014/04/22 00:00",
+        "submitterorganization":"JGI",
+        "refseq_category":"representative-genome",
+        "anomalouslist":[
+
+        ],
+        "propertylist":[
+          "full-genome-representation",
+          "latest",
+          "latest_genbank",
+          "latest_refseq",
+          "representative",
+          "species-default",
+          "wgs"
+        ],
+        "synonym":{
+          "genbank":"GCA_000315625.1",
+          "refseq":"GCF_000315625.1",
+          "similarity":"identical"
+        },
+        "sortorder":"2C50003156259898",
+        "meta":{
+          "Stats":{
+            "Stat":[
+              {
+                "_":"0",
+                "category":"alt_loci_count",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"0",
+                "category":"chromosome_count",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"5126",
+                "category":"contig_count",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"587",
+                "category":"contig_l50",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"40445",
+                "category":"contig_n50",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"0",
+                "category":"non_chromosome_replicon_count",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"0",
+                "category":"replicon_count",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"669",
+                "category":"scaffold_count",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"0",
+                "category":"scaffold_count",
+                "sequence_tag":"placed"
+              },
+              {
+                "_":"0",
+                "category":"scaffold_count",
+                "sequence_tag":"unlocalized"
+              },
+              {
+                "_":"669",
+                "category":"scaffold_count",
+                "sequence_tag":"unplaced"
+              },
+              {
+                "_":"52",
+                "category":"scaffold_l50",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"545808",
+                "category":"scaffold_n50",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"87145349",
+                "category":"total_length",
+                "sequence_tag":"all"
+              },
+              {
+                "_":"83457412",
+                "category":"ungapped_length",
+                "sequence_tag":"all"
+              }
+            ]
+          },
+          "FtpSites":{
+            "FtpPath":[
+              {
+                "_":"ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1",
+                "type":"GenBank"
+              },
+              {
+                "_":"ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCF_000315625.1_Guith1",
+                "type":"RefSeq"
+              }
+            ]
+          },
+          "assembly-level":"2",
+          "assembly-status":"Scaffold",
+          "representative-status":"representative-genome",
+          "submitter-organization":"JGI"
+        }
+      },
+      "urls": {
+        "rm" : {
+          "out" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_rm.out.gz",
+          "run" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_rm.run"
+        },
+        "uid" : "503988",
+        "stats" : {
+          "txt" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_assembly_stats.txt"
+        },
+        "protein" : {
+          "gpff" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_protein.gpff.gz",
+          "faa" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_protein.faa.gz"
+        },
+        "report" : {
+          "txt" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_assembly_report.txt"
+        },
+        "structure" : {
+          "dir" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_assembly_structure/"
+        },
+        "md5checksums" : {
+          "txt" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/md5checksums.txt"
+        },
+        "genomic" : {
+          "gbff" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_genomic.gbff.gz",
+          "gff" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_genomic.gff.gz",
+          "fna" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/GCA_000315625.1_Guith1_genomic.fna.gz"
+        },
+        "README" : {
+          "txt" : "http://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000315625.1_Guith1/README.txt"
+        }
+      }
     }
@@ -12,3 +215,3 @@ },
     "guillardia-theta": {
-      "download-list": [
+      "urls": [
         {"url":"http://ftp.ncbi.nlm.nih.gov/sra/sra-instant/reads/ByRun/sra/SRR/SRR070/SRR070672/SRR070672.sra","uid":"35523"},
@@ -15,0 +218,0 @@ {"url":"http://ftp.ncbi.nlm.nih.gov/sra/sra-instant/reads/ByRun/sra/SRR/SRR070/SRR070673/SRR070673.sra","uid":"35524"},

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

444520

increased by1.13%

Lines of code

1585

increased by15.95%

Worsened metrics

Dependency count

14

increased by7.69%

Dev dependency count

10

increased by42.86%

Dependency changes

• + Addedcheerio@^0.17.0

• + AddedCSSselect@0.4.1(transitive)

• + AddedCSSwhat@0.4.7(transitive)

• + Addedcheerio@0.17.0(transitive)

• + Addeddom-serializer@0.0.1(transitive)

• + Addeddomelementtype@1.1.3(transitive)

• + Addeddomhandler@2.2.1(transitive)

• + Addeddomutils@1.4.31.5.1(transitive)

• + Addedentities@1.0.01.1.2(transitive)

• + Addedhtmlparser2@3.7.3(transitive)

• + Addedlodash@2.4.2(transitive)