From d4564525a9efd14cf9562e8893837a873f5ed972 Mon Sep 17 00:00:00 2001
From: Peter van 't Hof <p.j.van_t_hof@lumc.nl>
Date: Tue, 17 Feb 2015 15:55:01 +0100
Subject: [PATCH] Move sickle summary methods to general sickle extension
---
.../lumc/sasc/biopet/extensions/Sickle.scala | 42 +++++++++++-
.../pipelines/flexiprep/Flexiprep.scala | 2 +-
.../biopet/pipelines/flexiprep/Sickle.scala | 66 -------------------
3 files changed, 42 insertions(+), 68 deletions(-)
delete mode 100644 public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Sickle.scala
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Sickle.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Sickle.scala
index abc517962..8a546534d 100644
--- a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Sickle.scala
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Sickle.scala
@@ -19,9 +19,13 @@ import java.io.File
import nl.lumc.sasc.biopet.core.BiopetCommandLineFunction
import nl.lumc.sasc.biopet.core.config.Configurable
+import nl.lumc.sasc.biopet.core.summary.Summarizable
import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
-class Sickle(val root: Configurable) extends BiopetCommandLineFunction {
+import scala.collection.mutable
+import scala.io.Source
+
+class Sickle(val root: Configurable) extends BiopetCommandLineFunction with Summarizable {
@Input(doc = "R1 input")
var input_R1: File = _
@@ -76,4 +80,40 @@ class Sickle(val root: Configurable) extends BiopetCommandLineFunction {
optional("--quiet", quiet) +
" > " + required(output_stats)
}
+
+ def summaryData: Map[String, Any] = {
+ val pairKept = """FastQ paired records kept: (\d*) \((\d*) pairs\)""".r
+ val singleKept = """FastQ single records kept: (\d*) \(from PE1: (\d*), from PE2: (\d*)\)""".r
+ val pairDiscarded = """FastQ paired records discarded: (\d*) \((\d*) pairs\)""".r
+ val singleDiscarded = """FastQ single records discarded: (\d*) \(from PE1: (\d*), from PE2: (\d*)\)""".r
+
+ var stats: mutable.Map[String, Int] = mutable.Map()
+
+ if (output_stats.exists) for (line <- Source.fromFile(output_stats).getLines) {
+ line match {
+ case pairKept(reads, pairs) => stats += ("num_paired_reads_kept" -> reads.toInt)
+ case singleKept(total, r1, r2) => {
+ stats += ("num_reads_kept_R1" -> r1.toInt)
+ stats += ("num_reads_kept_R2" -> r2.toInt)
+ }
+ case pairDiscarded(reads, pairs) => stats += ("num_paired_reads_discarded" -> reads.toInt)
+ case singleDiscarded(total, r1, r2) => {
+ stats += ("num_reads_discarded_R1" -> r1.toInt)
+ stats += ("num_reads_discarded_R2" -> r2.toInt)
+ }
+ case _ =>
+ }
+ }
+
+ stats.toMap ++ Map("version" -> getVersion)
+ }
+
+ override def resolveSummaryConflict(v1: Any, v2: Any, key: String): Any = {
+ (v1, v2) match {
+ case (v1: Int, v2: Int) => v1 + v2
+ case _ => v1
+ }
+ }
+
+ def summaryFiles: Map[String, File] = Map()
}
diff --git a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala b/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala
index 785efc483..91defcc85 100644
--- a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala
+++ b/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala
@@ -21,7 +21,7 @@ import org.broadinstitute.gatk.utils.commandline.{ Input, Argument }
import nl.lumc.sasc.biopet.core.{ SampleLibraryTag, BiopetQScript, PipelineCommand }
import nl.lumc.sasc.biopet.core.config.Configurable
-import nl.lumc.sasc.biopet.extensions.{ Gzip, Pbzip2, Md5sum, Zcat, Seqstat }
+import nl.lumc.sasc.biopet.extensions._
import nl.lumc.sasc.biopet.tools.FastqSync
class Flexiprep(val root: Configurable) extends QScript with SummaryQScript with SampleLibraryTag {
diff --git a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Sickle.scala b/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Sickle.scala
deleted file mode 100644
index ec452f187..000000000
--- a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Sickle.scala
+++ /dev/null
@@ -1,66 +0,0 @@
-/**
- * Biopet is built on top of GATK Queue for building bioinformatic
- * pipelines. It is mainly intended to support LUMC SHARK cluster which is running
- * SGE. But other types of HPC that are supported by GATK Queue (such as PBS)
- * should also be able to execute Biopet tools and pipelines.
- *
- * Copyright 2014 Sequencing Analysis Support Core - Leiden University Medical Center
- *
- * Contact us at: sasc@lumc.nl
- *
- * A dual licensing mode is applied. The source code within this project that are
- * not part of GATK Queue is freely available for non-commercial use under an AGPL
- * license; For commercial users or users who do not want to follow the AGPL
- * license, please contact us to obtain a separate license.
- */
-package nl.lumc.sasc.biopet.pipelines.flexiprep
-
-import java.io.File
-
-import nl.lumc.sasc.biopet.core.config.Configurable
-
-import argonaut._, Argonaut._
-import nl.lumc.sasc.biopet.core.summary.Summarizable
-import scalaz._, Scalaz._
-
-import scala.io.Source
-import scala.collection.mutable
-
-class Sickle(root: Configurable) extends nl.lumc.sasc.biopet.extensions.Sickle(root) with Summarizable {
-
- def summaryData: Map[String, Any] = {
- val pairKept = """FastQ paired records kept: (\d*) \((\d*) pairs\)""".r
- val singleKept = """FastQ single records kept: (\d*) \(from PE1: (\d*), from PE2: (\d*)\)""".r
- val pairDiscarded = """FastQ paired records discarded: (\d*) \((\d*) pairs\)""".r
- val singleDiscarded = """FastQ single records discarded: (\d*) \(from PE1: (\d*), from PE2: (\d*)\)""".r
-
- var stats: mutable.Map[String, Int] = mutable.Map()
-
- if (output_stats.exists) for (line <- Source.fromFile(output_stats).getLines) {
- line match {
- case pairKept(reads, pairs) => stats += ("num_paired_reads_kept" -> reads.toInt)
- case singleKept(total, r1, r2) => {
- stats += ("num_reads_kept_R1" -> r1.toInt)
- stats += ("num_reads_kept_R2" -> r2.toInt)
- }
- case pairDiscarded(reads, pairs) => stats += ("num_paired_reads_discarded" -> reads.toInt)
- case singleDiscarded(total, r1, r2) => {
- stats += ("num_reads_discarded_R1" -> r1.toInt)
- stats += ("num_reads_discarded_R2" -> r2.toInt)
- }
- case _ =>
- }
- }
-
- stats.toMap ++ Map("version" -> getVersion)
- }
-
- override def resolveSummaryConflict(v1: Any, v2: Any, key: String): Any = {
- (v1, v2) match {
- case (v1: Int, v2: Int) => v1 + v2
- case _ => v1
- }
- }
-
- def summaryFiles: Map[String, File] = Map()
-}
--
GitLab