From 734801f6c59a3e63a759e378e2dcf0ec9a19cdc0 Mon Sep 17 00:00:00 2001
From: sajvanderzeeuw <s.a.j.van_der_zeeuw@lumc.nl>
Date: Fri, 5 Feb 2016 10:48:34 +0100
Subject: [PATCH] add DNA base check (testing still needed)
---
.../sasc/biopet/tools/CheckFastqPairs.scala | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/CheckFastqPairs.scala b/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/CheckFastqPairs.scala
index d3b45590f..e4c78dd32 100644
--- a/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/CheckFastqPairs.scala
+++ b/public/biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/CheckFastqPairs.scala
@@ -55,15 +55,30 @@ object CheckFastqPairs extends ToolCommand {
//Getting R2 record, None if it's single end
val recordR2 = readFq2.map(_.next())
- //Here we check if the readnames of both files are concordant
+ //Here we check if the readnames of both files are concordant, and if the sequence content are correct DNA/RNA sequences
recordR2 match {
case Some(recordR2) => // Paired End
val readHeader = recordR1.getReadHeader
val readHeader2 = recordR2.getReadHeader
+ val readSeq = recordR1.getReadString
+ val readSeq2 = recordR2.getReadString
val id1 = readHeader.takeWhile(_ != ' ')
val id2 = readHeader2.takeWhile(_ != ' ')
- if (counter % 1e5 == 0) logger.info(counter + " reads processed")
+ if (counter % 1e4 == 0) logger.info(counter + " reads processed")
+
+
+ val allowedBases = """([actgnACTGN+]+)""".r
+
+ val validBases: Boolean = readSeq match {
+ case allowedBases(m) => true
+ case _ => throw new IllegalStateException(s"Non IUPAC symbols identified '${(counter*4)-3}'")
+ }
+
+ val validBases2: Boolean = readSeq2 match {
+ case allowedBases(m) => true
+ case _ => throw new IllegalStateException(s"Non IUPAC symbols identified '${(counter*4)-3}'")
+ }
if (id1 == id2){
--
GitLab