From db1a67557fad2f4dbee6998c71e33b50f0770396 Mon Sep 17 00:00:00 2001
From: Peter van 't Hof <p.j.van_t_hof@lumc.nl>
Date: Tue, 29 Mar 2016 13:32:26 +0200
Subject: [PATCH] Fixed sorting issue
---
.../nl/lumc/sasc/biopet/tools/GensToVcf.scala | 15 +++++++++++++--
1 file changed, 13 insertions(+), 2 deletions(-)
diff --git a/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/GensToVcf.scala b/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/GensToVcf.scala
index d02483747..143228366 100644
--- a/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/GensToVcf.scala
+++ b/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/GensToVcf.scala
@@ -83,7 +83,7 @@ object GensToVcf extends ToolCommand {
lazy val fastaFile = ReferenceSequenceFileFactory.getReferenceSequenceFile(cmdArgs.referenceFasta, true, true)
case class Line(genotype: String, info: Option[String])
- def lineIt: Iterator[Line] = {
+ val lineIt: Iterator[Line] = {
val it = infoIt match {
case Some(x) => genotypeIt.zip(x).map(x => Line(x._1, Some(x._2)))
case _ => genotypeIt.map(x => Line(x, None))
@@ -92,13 +92,20 @@ object GensToVcf extends ToolCommand {
if (cmdArgs.sortInput) {
logger.info("Start Sorting input files")
val list = it.toList
- val pos = list.map(_.genotype.split(" ")(2).toInt)
+ val pos = list.map{ line =>
+ val values = line.genotype.split(" ")
+ val p = values(2).toInt
+ val alt = values(4)
+ if (alt == "-") p - 1
+ else p
+ }
list.zip(pos).sortBy(_._2).map(_._1).toIterator
}
else it
}
logger.info("Start processing genotypes")
+ var count = 0L
for (line <- lineIt) {
val genotypeValues = line.genotype.split(" ")
val (start, end, ref, alt) = {
@@ -141,8 +148,12 @@ object GensToVcf extends ToolCommand {
val id = genotypeValues(1)
if (id.startsWith(cmdArgs.contig + ":")) writer.add(builder.make())
else writer.add(builder.id(id).make())
+ count += 1
+ if (count % 10000 == 0) logger.info(s"$count lines processed")
}
+ logger.info(s"$count lines processed")
+
writer.close()
logger.info("Done")
--
GitLab