From 38d3970ed6fdb323e31fb19e68c7cea8ca35aa07 Mon Sep 17 00:00:00 2001
From: Peter van 't Hof <p.j.van_t_hof@lumc.nl>
Date: Sun, 23 Aug 2015 11:03:35 +0200
Subject: [PATCH] Added lengths
---
.../scala/nl/lumc/sasc/biopet/tools/SquishBed.scala | 11 +++++++++++
.../lumc/sasc/biopet/utils/intervals/BedRecord.scala | 2 ++
.../sasc/biopet/utils/intervals/BedRecordList.scala | 2 ++
3 files changed, 15 insertions(+)
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 971be2f46..7a424f510 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
@@ -38,7 +38,18 @@ object SquishBed extends ToolCommand {
logger.info("Start")
val records = BedRecordList.fromFile(cmdArgs.input)
+ val length = records.length
+ val refLength = BedRecordList.combineOverlap(records).length
+ logger.info(s"Total bases: $length")
+ logger.info(s"Total bases on reference: $refLength")
+ logger.info("Start squishing")
val squishBed = records.squishBed(cmdArgs.strandSensitive).sort
+ logger.info("Done squishing")
+ val squishLength = squishBed.length
+ val squishRefLength = BedRecordList.combineOverlap(squishBed).length
+ logger.info(s"Total bases left: $squishLength")
+ logger.info(s"Total bases left on reference: $squishRefLength")
+ logger.info(s"Total bases removed from ref: ${refLength - squishRefLength}")
squishBed.writeToFile(cmdArgs.output)
logger.info("Done")
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecord.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecord.scala
index a7e346fb6..543e41509 100644
--- a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecord.scala
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/utils/intervals/BedRecord.scala
@@ -24,6 +24,8 @@ case class BedRecord(chr: String,
else _originals
}
+ def length = end - start + 1
+
lazy val exons = if (blockCount.isDefined && blockSizes.length > 0 && blockStarts.length > 0) {
Some(BedRecordList.fromList(for (i <- 0 to blockCount.get) yield {
val exonNumber = strand match {
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 20ee5cd81..ed2c873d3 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
@@ -25,6 +25,8 @@ class BedRecordList(val chrRecords: Map[String, List[BedRecord]]) {
.dropWhile(_.end < record.start)
.takeWhile(_.start <= record.end)
+ def length = allRecords.foldLeft(0L)((a,b) => a + b.length)
+
def squishBed(strandSensitive: Boolean = true) = BedRecordList.fromList {
(for ((chr, records) <- sort.chrRecords; record <- records) yield {
val overlaps = overlapWith(record)
--
GitLab