WDL to Nextflow

Workflow general structure

WDL

Nextflow

version <VERSION> #Processes imported from other diagrams import “diagram_1” as ext_1 … #Processes (or tasks) defined in diagram task task_1 { ##Task body } … #Main workflow workflow mainWorkflow { input { <Type> p_1 … } #Workflow body call task_1 as call_1 { input_1 = value_1, … } … expression_1 … scatter_1 (element in array) { #Scatter body } … output{ <Type> p_1 = <Expression> … } }

//Groovy functions defined in special nextflow document include { basename; sub; length; ... } from './biouml_function.nf' //Processes imported from other diagrams include { ext_process_1 as ext_1 } from './diagram_1' … //Default Global workflow inputs params.p_1 … //Processes defined in diagram process_1 {…} … //Main workflow workflow mainWorkflow { take: p_1 … main: call_1 {…} expression_1 … emit: result_1 = expression_1 … } //Entry point workflow workflow { mainWorkflow( params.p_1, …, params.p_k ) }

Task and Process general structure

WDL	Nextflow
task task_name { input { <Type> input_1 … <Type> optional_input_1 = <Expression> … } <Type> variable_1 = <Expression> … command { <Executable script> } output { <Type> output_1 = <Expression> } runtime { <Runtime properties> } }	process p_1 { input : <Type> input_1 … publishDir "p_1", mode: 'copy' fair true script: variable_1 = <Expression> … """ <Executable script> """ output: path output_1, emit: out_1 … }

Type mapping

#	WDL	Nextflow
	File, Array[File]	path
	Boolean, String, Integer, Float, Array[Boolean], Array[String], Array[Integer], Array[Float]	val
	Map, Struct, Object	val

Imports

#	Description	WDL	Nextflow
	Import task from other script	import “script” as ext ... call ext.task1 as t { … }	include { task1 as t } from './script ... t( … )
	Call imported task twice	import “script” as ext … call ext.task1 as t1 { … } call ext.task1 as t2 { … }	include { task1 as t1 } from './script include { task1 as t2 } from './script … t1( … ) t2( … )
	Import external workflow	import “script” as ext call ext.mainWorkflow as alias { … }	include { mainWorkflow as alias } from './script

Expressions

#	Description	WDL	Nextflow
	Workflow variable	~{var_name}	${var_name}
	Bash variable	${var_name}	\${var_name}
	Cycle in script	for file in ~{sep=" " files}	for file in ${ files.join() }
	Ternary operator	~{ if ... then ... else ... } if ... then ... else ...	... ? ... : ...
	Default value	~{ default="default_value" var }	${ var ?: "default_value" }
	Glob in result	glob("output/file_*.txt")	"output/file_*.txt"
	Predefined functions	basename( file ) sub( input, pattern, replacement ) length( array ) ...	include { basename; sub; length; ... } from './biouml_function.nf' basename( file ) sub( input, pattern, replacement ) length(array) ...
	Array element access	x = array[i]	//Use Groovy function, array can be channel x = get( array, i )

Task elements

#	Description	WDL	Nextflow
	Task metadata	meta { description: "text" }	Not supported
	Parameters metadata	parameter_meta { input_vcf: { help: “help text” description: “text” suggestions: [“v1”,”v2”] } }	Not supported
	Private declarations	task task1 { … <Type> variable = <Value> … command { ...	process task_name … script variable = <Value> """ …
	Runtime options	runtime { maxRetries: 3 memory: "N GB" preemptible: 1 disks: "local-disk ~{disk_size} HDD" cpu: threads docker: "image name" }	maxRetries 3 memory "4*4 GB" cpus 4 container 'image_name'
	Optional inputs	input { Sting mandatory String opt = “Default” }	input: val mandatory val opt … script: opt = getDefault(opt, “Default”) """ … """

Workflow

#	Description	WDL	Nextflow
	Two consequent steps	call task1 as call1 { input: arg1 = val1 } call task2 as call2 { input: arg1 = call1.result, arg2 = val2 }	result_task1 = task1( val1 ) task2( result_task1.result, val2 )
	Calling task with optional inputs	call task1 { mandatory = value #optional = value2 }	task1( value , “NO_VALUE”)
	Multiple calls of task defined in the same document	task task1 {…} call task1 as t1 { } call task1 as t2 { }	Not supported
	Expressions inside if	if (x > 2) { Int y = x*2 }	y = x.map{ (it>2)? it*2 : null }

Cycles

#	Description	WDL	Nextflow
	Simple cycle	scatter ( file in files ) { call task1 { input: arg1 = file } }	file = files task1( file )
	Cycle in range	scatter ( i in range(length(С)) ) { call task1 { input: arg1 = i } }	//Use Groove functions i = range( length(С) ) task1( i )
	Expressions in cycle	scatter ( i in array ) { Int j = i*2 Int k = i*3 }	j = toChannel( array ).map { i -> i*2 } k = toChannel( array ).map { i -> i*3 }
	Nested cycles	scatter ( i in array ) { scatter ( j in array2 ) { Array[Int] pair = [ i,j ] } }	pair = toChannel( array ).combine(toChannel(array2)).map{ i,j -> [i,j] }

Supplementary functions

#	Function	Description	Code
	get(arr, i)	Returns element of arr with index i, arr may be Array or Channel.	TBA
	getDefault(var, Default)	Returns var or Default if var is not defined.	TBA
	toChannel(arr)	Creates channel from arr, if arr is channel already does nothing.	TBA

WDL functions implemented in Groovy

#	Function	TBA
	basename(path)	TBA
	sub(input, pattern, replacement)	TBA
	ceil	TBA
	length	TBA
	range	TBA
	select_first	TBA
	select_all	TBA
	defined	TBA
	read_string	TBA
	read_int	TBA
	read_float	TBA

Not supported yet

#	Name	Description
	Comments	Comments.