add kallisto quantification

workflow/build.gradle.kts

@@ -9,7 +9,7 @@ plugins {
 }
 
 group = "com.genomealmanac.rnaseq"
-version = "1.0.0"
+version = "1.1.0"
 val artifactID = "rnaseq-workflow"
 
 repositories {

workflow/src/main/kotlin/workflow/task/Kallisto.kt

@@ -0,0 +1,54 @@
+package workflow.task
+
+import krews.core.WorkflowBuilder
+import krews.core.*
+import krews.file.File
+import krews.file.OutputFile
+import org.reactivestreams.Publisher
+
+data class KallistoParams (
+    val index: File,
+    val fragmentLength: Int? = 100,
+    val sdFragmentLength: Int? = 10,
+    val strandedness: String = "unstranded",
+    val cores: Int = 1
+)
+
+data class KallistoInput (
+    val r1: File,
+    val r2: File? = null,
+    val name: String
+)
+
+data class KallistoOutput (
+    val name: String,
+    val quantifications: File
+)
+
+fun WorkflowBuilder.kallistoTask(name: String, i: Publisher<KallistoInput>)
+  = this.task<KallistoInput, KallistoOutput>(name, i) {
+
+    val params = taskParams<KallistoParams>()
+    dockerImage = "docker.pkg.github.com/krews-community/rnaseq-kallisto-task/rnaseq-kallisto:1.1.0"
+    val prefix = "${input.name}"
+
+    output = KallistoOutput(
+        name = input.name,
+        quantifications = OutputFile("${prefix}.abundance.tsv")
+    )
+
+    command = """
+        java -XX:+UnlockExperimentalVMOptions -XX:+UseCGroupMemoryLimitForHeap -XX:MaxRAMFraction=1 \
+            -jar /app/kallisto.jar quant \
+                --index ${params.index.dockerPath} \
+                --output-directory ${outputsDir} \
+                --cores ${params.cores} \
+                --strandedness ${params.strandedness} \
+                ${ if (input.r2 === null) "--fragment-length ${params.fragmentLength}" else "" } \
+                ${ if (input.r2 === null) "--sd-fragment-length ${params.sdFragmentLength}" else "" } \
+                --output-prefix ${input.name} \
+                --r1 ${input.r1.dockerPath} \
+                ${ if (input.r2 !== null) "--r2  ${input.r2!!.dockerPath}" else "" }
+    """
+
+}

workflow/src/main/kotlin/workflow/workflow/RnaSeqWorkflow.kt

@@ -45,6 +45,12 @@ val rnaSeqWorkflow = workflow("encode-rnaseq-workflow") {
     }.toFlux()
     val mergeFastqTask = MergeFastqTask("merge", mergeFastqInput)
 
+    /* run kallisto quant for experiments starting from FASTQs */
+    val kallistoInput = mergeFastqTask.map {
+        KallistoInput(it.mergedFileR1, it.mergedFileR2, it.repName)
+    }
+    kallistoTask("kallisto", kallistoInput)
+
     /* create task to align reads for experiments starting from FASTQs */
     val starInput = mergeFastqTask
         .map { AlignerInput(it.mergedFileR1, it.mergedFileR2, it.repName, it.pairedEnd) }        

workflow/src/test/kotlin/workflow/AppTests.kt

@@ -83,6 +83,15 @@ class AppTests {
                 local-path = "$testDir/test.index.tar.gz"
             }
         }
+
+        task.kallisto.params {
+            index = {
+                -type = "krews.file.LocalInputFile"
+                local-path = "$testDir/hg38.chrM.kallisto.idx"
+            }
+            fragment-length = 100
+            sd-fragment-length = 10
+        }
     """.trimIndent()
 
     @BeforeAll
@@ -104,6 +113,9 @@ class AppTests {
         /* FASTQ merging task */
         assertMD5(outputsDir.resolve("testrep1.merged.r1.fastq.gz"), "6b0d6ea05ab99a717a247e0b6ace5228")
 
+        /* kallisto task */
+        assertMD5(outputsDir.resolve("testrep1.abundance.tsv"), "5f9faa00904649814f7e030790164e6a")
+
         /* STAR task */
         assertMD5(outputsDir.resolve("testrep1_anno.bam"), "1644cde899e6e71de60b669855e315bc")
         assertMD5(outputsDir.resolve("testrep1_anno_flagstat.txt"), "87c78efa9ad0a76dec66ab5c8e8e7754")