-
Notifications
You must be signed in to change notification settings - Fork 309
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Reimplement FASTA sequence and slice converters for performance.
- Loading branch information
Showing
7 changed files
with
346 additions
and
535 deletions.
There are no files selected for viewing
301 changes: 301 additions & 0 deletions
301
adam-core/src/main/scala/org/bdgenomics/adam/converters/FastaConverters.scala
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,301 @@ | ||
/** | ||
* Licensed to Big Data Genomics (BDG) under one | ||
* or more contributor license agreements. See the NOTICE file | ||
* distributed with this work for additional information | ||
* regarding copyright ownership. The BDG licenses this file | ||
* to you under the Apache License, Version 2.0 (the | ||
* "License"); you may not use this file except in compliance | ||
* with the License. You may obtain a copy of the License at | ||
* | ||
* http://www.apache.org/licenses/LICENSE-2.0 | ||
* | ||
* Unless required by applicable law or agreed to in writing, software | ||
* distributed under the License is distributed on an "AS IS" BASIS, | ||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
* See the License for the specific language governing permissions and | ||
* limitations under the License. | ||
*/ | ||
package org.bdgenomics.adam.converters | ||
|
||
import org.apache.spark.rdd.RDD | ||
import org.apache.spark.broadcast.Broadcast | ||
import org.bdgenomics.formats.avro.{ Alphabet, Sequence, Slice } | ||
import scala.collection.immutable.SortedMap | ||
import scala.collection.mutable.MutableList | ||
|
||
/** | ||
* FASTA format description line. | ||
* | ||
* @param index Index of the description line. | ||
* @param name Name parsed from the description line. | ||
* @param description Description parsed from the description line. | ||
*/ | ||
private[adam] case class FastaDescriptionLine( | ||
index: Long = -1L, | ||
name: Option[String] = None, | ||
description: Option[String] = None) { | ||
} | ||
|
||
/** | ||
* FASTA format converters. | ||
*/ | ||
private[converters] object FastaConverters { | ||
/** | ||
* Regular expression to identify valid FASTA format lines, primarly of use when | ||
* reading FASTA formatted sequences from GFF3 files. Nucleotide symbols, amino acid | ||
* symbols, and ambiguity code symbols cover a-z. Also valid are whitespace, gap | ||
* symbol -, and translation stop *. | ||
*/ | ||
private val fastaRegex = "^[a-zA-Z-*\\s]+$".r | ||
|
||
/** | ||
* Return true if the specified line is a description line. | ||
* | ||
* @param line FASTA format line. | ||
* @return Return true if the specified line is a description line. | ||
*/ | ||
def isDescriptionLine(line: String): Boolean = { | ||
line.startsWith(">") | ||
} | ||
|
||
/** | ||
* Return true if the specified line is valid FASTA format. | ||
* | ||
* @param line FASTA format line. | ||
* @return Return true if the specified line is valid FASTA format. | ||
*/ | ||
def isFastaLine(line: String): Boolean = { | ||
isDescriptionLine(line) || fastaRegex.pattern.matcher(line).matches() | ||
} | ||
|
||
/** | ||
* Clean up a sequence by stripping asterisks at the end of the sequence. | ||
* | ||
* To be consistent with a legacy database, some FASTA sequences end in a "*" | ||
* suffix. This method strips that suffix from the end of the sequence. | ||
* | ||
* @param sequence The sequence to clean. | ||
* @return Sequence minus "*" suffix. | ||
*/ | ||
def cleanSequence(sequence: String): String = { | ||
sequence.stripSuffix("*") | ||
} | ||
|
||
/** | ||
* Parse the specified line into a description line. | ||
* | ||
* @param index Index of the line to parse. | ||
* @param optLine Line to parse, if any. | ||
* @return Return the specified line parsed into a description line. | ||
*/ | ||
def parseDescriptionLine(index: Long, optLine: Option[String]): FastaDescriptionLine = { | ||
optLine.fold { | ||
require(index == -1L, "Cannot have a headerless line in a file with more than one sequence.") | ||
FastaDescriptionLine(index, None, None) | ||
} { | ||
(line) => | ||
|
||
// fasta description line splits on whitespace | ||
val splitIndex = line.indexWhere(c => c.isWhitespace) | ||
if (splitIndex >= 0) { | ||
val split = line.splitAt(splitIndex) | ||
|
||
// is this description metadata or not? if it is metadata, it will contain "|" | ||
if (split._1.contains('|')) { | ||
FastaDescriptionLine(index, None, Some(line.stripPrefix(">").trim)) | ||
} else { | ||
val name: String = split._1.stripPrefix(">").trim | ||
val description: String = split._2.trim | ||
|
||
FastaDescriptionLine(index, Some(name), Some(description)) | ||
} | ||
} else { | ||
FastaDescriptionLine(index, Some(line.stripPrefix(">").trim), None) | ||
} | ||
} | ||
} | ||
|
||
/** | ||
* Filter the specified lines to FASTA format lines. | ||
* | ||
* @param rdd Lines keyed by line number. | ||
* @return Return the specified lines filtered to FASTA format lines. | ||
*/ | ||
def filterToFastaLines(rdd: RDD[(Long, String)]): RDD[(Long, String)] = { | ||
rdd | ||
.map(kv => (kv._1, kv._2.trim())) | ||
.filter((kv: (Long, String)) => isFastaLine(kv._2)) | ||
} | ||
|
||
/** | ||
* Collect description lines from the specified lines into a sorted map suitable | ||
* for broadcast. | ||
* | ||
* @param rdd Lines keyed by line number. | ||
* @return Return description lines from the specified lines collected into a sorted | ||
* map suitable for broadcast. | ||
*/ | ||
def collectDescriptionLines(rdd: RDD[(Long, String)]): SortedMap[Long, FastaDescriptionLine] = { | ||
val descriptionLines: SortedMap[Long, FastaDescriptionLine] = SortedMap() | ||
|
||
descriptionLines ++ rdd // Scala 2.13+ only SortedMap.from( | ||
.filter(kv => isDescriptionLine(kv._2)) | ||
.map(kv => (kv._1, parseDescriptionLine(kv._1, Some(kv._2)))) | ||
.sortByKey() // not strictly necessary, slightly slower without | ||
.collect() | ||
} | ||
|
||
/** | ||
* Group sequence lines by description line index. | ||
* | ||
* @param rdd FASTA format sequence lines keyed by line number. | ||
* @param broadcast Sorted map of description lines keyed by line number. | ||
* @return Return sequence lines grouped by description line index. | ||
*/ | ||
def groupByDescriptionLineIndex(rdd: RDD[(Long, String)], broadcast: Broadcast[SortedMap[Long, FastaDescriptionLine]]): RDD[(Long, Iterable[(Long, String)])] = { | ||
rdd | ||
.filter(kv => !isDescriptionLine(kv._2)) | ||
//.keyBy(kv => broadcast.value.maxBefore(kv._1).getOrElse(-1L)) Scala 2.13+ only | ||
.keyBy(kv => broadcast.value.until(kv._1).lastOption.getOrElse((-1L, Nil))._1) | ||
.groupByKey() | ||
} | ||
} | ||
|
||
/** | ||
* Convert FASTA format into Sequences. | ||
*/ | ||
private[adam] object FastaSequenceConverter { | ||
import FastaConverters._ | ||
|
||
/** | ||
* Convert FASTA format into Sequences. | ||
* | ||
* @param alphabet Alphabet for the sequences. | ||
* @param rdd RDD of FASTA format as (Long, String) tuples. | ||
* @return Return an RDD of Sequences converted from FASTA format. | ||
*/ | ||
def apply(alphabet: Alphabet, rdd: RDD[(Long, String)]): RDD[Sequence] = { | ||
|
||
val filtered = filterToFastaLines(rdd) | ||
val descriptionLines = collectDescriptionLines(filtered) | ||
val broadcast = rdd.context.broadcast(descriptionLines) | ||
val sequenceLines = groupByDescriptionLineIndex(filtered, broadcast) | ||
|
||
sequenceLines.flatMap { | ||
case (index: Long, lines) => | ||
|
||
val descriptionLine = broadcast.value.getOrElse(index, FastaDescriptionLine()) | ||
|
||
val sequence = lines.toSeq.sortBy(_._1).map(kv => cleanSequence(kv._2)).mkString | ||
val length = sequence.length().toLong | ||
|
||
val builder = Sequence.newBuilder() | ||
.setAlphabet(alphabet) | ||
.setSequence(sequence) | ||
.setLength(length) | ||
|
||
// map over optional fields | ||
descriptionLine.name.foreach(builder.setName(_)) | ||
descriptionLine.description.foreach(builder.setDescription(_)) | ||
|
||
Seq(builder.build) | ||
} | ||
} | ||
} | ||
|
||
/** | ||
* Convert FASTA format into Slices. | ||
*/ | ||
private[adam] object FastaSliceConverter { | ||
import FastaConverters._ | ||
|
||
/** | ||
* Convert FASTA format into Slices. | ||
* | ||
* @param rdd RDD of FASTA format as (Long, String) tuples. | ||
* @param maximumLength Maximum length in bases for each Slice. | ||
* @return Return an RDD of Slices converted from FASTA format. | ||
*/ | ||
def apply(rdd: RDD[(Long, String)], maximumLength: Long): RDD[Slice] = { | ||
|
||
val filtered = filterToFastaLines(rdd) | ||
val descriptionLines = collectDescriptionLines(filtered) | ||
val broadcast = rdd.context.broadcast(descriptionLines) | ||
val sequenceLines = groupByDescriptionLineIndex(filtered, broadcast) | ||
|
||
sequenceLines.flatMap { | ||
case (index: Long, lines) => | ||
|
||
val descriptionLine = broadcast.value.getOrElse(index, FastaDescriptionLine()) | ||
|
||
val sequences = lines.toSeq.sortBy(_._1).map(kv => cleanSequence(kv._2)) | ||
val length = sequences.map(_.length).sum | ||
|
||
val sequenceSlices = slice(sequences, maximumLength) | ||
val sliceCount = sequenceSlices.length | ||
|
||
val slices = sequenceSlices.zipWithIndex | ||
.map(si => { | ||
val (bases, index) = si | ||
|
||
val builder = Slice.newBuilder() | ||
.setAlphabet(Alphabet.DNA) | ||
.setSequence(bases) | ||
.setIndex(index) | ||
.setStart(index * maximumLength) | ||
.setEnd(index * maximumLength + bases.length) | ||
.setSlices(sliceCount) | ||
.setLength(bases.length.toLong) | ||
.setTotalLength(length.toLong) | ||
|
||
// map over optional fields | ||
descriptionLine.name.foreach(builder.setName(_)) | ||
descriptionLine.description.foreach(builder.setDescription(_)) | ||
|
||
builder.build | ||
}) | ||
|
||
slices | ||
} | ||
} | ||
|
||
/** | ||
* Slice the specified sequence strings per the specified maximum length. | ||
* | ||
* @param sequences Sequence strings to slice. | ||
* @param maximumLength Maximum length in bases for each slice. | ||
* @return Return the specified sequence strings sliced per the specified maximum length. | ||
*/ | ||
private def slice(sequences: Seq[String], maximumLength: Long): Seq[String] = { | ||
// internal "fsm" variables | ||
var sequence: StringBuilder = new StringBuilder | ||
var sequenceSeq: MutableList[String] = MutableList() | ||
|
||
/** | ||
* Adds a string slice to our accumulator. If this string slice causes the accumulator | ||
* to grow longer than our max slice size, we split the accumulator and add it to the end | ||
* of our list of slice. | ||
* | ||
* @param seq Slice string to add. | ||
*/ | ||
def addSlice(seq: String) { | ||
sequence.append(seq) | ||
|
||
while (sequence.length > maximumLength) { | ||
sequenceSeq += sequence.take(maximumLength.toInt).toString() | ||
sequence = sequence.drop(maximumLength.toInt) | ||
} | ||
} | ||
|
||
// run addFragment on all slices | ||
sequences.foreach(addSlice) | ||
|
||
// if we still have a remaining sequence that is not part of a slice, add it | ||
if (sequence.nonEmpty) { | ||
sequenceSeq += sequence.toString() | ||
} | ||
|
||
// return our slices | ||
sequenceSeq | ||
} | ||
} |
Oops, something went wrong.