Krews, short for "Kotlin Reactive Workflows" is a framework for creating scalable and reproducible scientific data pipelines using Docker Containers piped together with Project Reactor, a mature functional reactive programming library.

Workflows are written using a Kotlin DSL in plain old Kotlin projects, meaning you get all the benefits of a modern, type-safe, functional language with fantastic tooling.

Full Documentation can be found here

The fastest way to get started is to clone the krews-boilerplate. This is not only quick, but is set up using what we consider to be best practices.

The boilerplate README.md contains instructions on what you'll need to replace, as well as instructions on running and building.

fun main(args: Array<String>) = run(sampleWorkflow, args)

val sampleWorkflow = workflow("sample-workflow") {

    // Reactive "Flux" list object for the numbers 1 to 5
    val range = (1..5).toFlux()
    
    task<Int, File>("base64", range) {
        dockerImage = "alpine:3.9"
        output = OutputFile("base64/$input.b64")
        command =
            """
            mkdir -p /data/base64
            echo "Hello World $input!" | base64 > /data/base64/$input.b64
            """
    }
}

Configuration

working-dir = /data/sample-workflow

Run command

java -jar my-app.jar --on local --config path/to/my-config.conf

• Google Cloud Life Sciences
• Slurm
• Local Docker (Single Machine)

dependencies {
    compile("io.krews", "krews", "0.10.10")
}

<dependency>
  <groupId>io.krews</groupId>
  <artifactId>krews</artifactId>
  <version>0.10.10</version>
</dependency>