From 3483a1be6cab2482d328ee7ebd45a932db514dc7 Mon Sep 17 00:00:00 2001
From: Peter van 't Hof <p.j.van_t_hof@lumc.nl>
Date: Sun, 23 Aug 2015 10:48:38 +0200
Subject: [PATCH] cached sorting
---
.../nl/lumc/sasc/biopet/tools/SquishBed.scala | 2 +-
.../utils/intervals/BedRecordList.scala | 25 ++++++++++---------
2 files changed, 14 insertions(+), 13 deletions(-)
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/tools/SquishBed.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/tools/SquishBed.scala
index 76bca165a..971be2f46 100644
--- a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/tools/SquishBed.scala
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/tools/SquishBed.scala
@@ -37,7 +37,7 @@ object SquishBed extends ToolCommand {
logger.info("Start")
- val records = BedRecordList.fromFile(cmdArgs.input).sort
+ val records = BedRecordList.fromFile(cmdArgs.input)
val squishBed = records.squishBed(cmdArgs.strandSensitive).sort
squishBed.writeToFile(cmdArgs.output)
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecordList.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecordList.scala
index a0deb868b..20ee5cd81 100644
--- a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecordList.scala
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecordList.scala
@@ -13,21 +13,20 @@ import nl.lumc.sasc.biopet.core.Logging
class BedRecordList(val chrRecords: Map[String, List[BedRecord]]) {
def allRecords = for (chr <- chrRecords; record <- chr._2) yield record
- def sort = new BedRecordList(chrRecords.map(x => x._1 -> x._2.sortWith((a, b) => a.start < b.start)))
-
- lazy val isSorted = {
- val sorted = this.sort
- sorted.chrRecords.forall(x => x._2 == chrRecords(x._1))
+ lazy val sort = {
+ val sorted = new BedRecordList(chrRecords.map(x => x._1 -> x._2.sortWith((a, b) => a.start < b.start)))
+ if (sorted.chrRecords.forall(x => x._2 == chrRecords(x._1))) this else sorted
}
- def overlapWith(record: BedRecord) = (if (isSorted) this else sort).chrRecords
+ lazy val isSorted = sort.hashCode() == this.hashCode() || sort.chrRecords.forall(x => x._2 == chrRecords(x._1))
+
+ def overlapWith(record: BedRecord) = sort.chrRecords
.getOrElse(record.chr, Nil)
.dropWhile(_.end < record.start)
.takeWhile(_.start <= record.end)
def squishBed(strandSensitive: Boolean = true) = BedRecordList.fromList {
- if (!isSorted) Logging.logger.warn("Running squish bed method on a unsorted bed file may not work correctly")
- (for ((chr, records) <- chrRecords; record <- records) yield {
+ (for ((chr, records) <- sort.chrRecords; record <- records) yield {
val overlaps = overlapWith(record)
.filterNot(strandSensitive && _.strand != record.strand)
.filterNot(_.name == record.name)
@@ -60,10 +59,12 @@ object BedRecordList {
def fromList(records: Traversable[BedRecord]): BedRecordList = fromList(records.toIterator)
def fromList(records: TraversableOnce[BedRecord]): BedRecordList = {
- val map = mutable.Map[String, List[BedRecord]]()
- for (record <- records)
- map += record.chr -> (record :: map.getOrElse(record.chr, List()))
- new BedRecordList(map.toMap)
+ val map = mutable.Map[String, ListBuffer[BedRecord]]()
+ for (record <- records) {
+ if (!map.contains(record.chr)) map += record.chr -> ListBuffer()
+ map(record.chr) += record
+ }
+ new BedRecordList(map.toMap.map(m => m._1 -> m._2.toList))
}
def fromFile(bedFile: File) = {
--
GitLab