NextflowWorkbench: Reproducible and Reusable Workflows for Beginners and Experts

Computational workflows and pipelines are often created to automate series of processing steps. For instance, workflows enable one to standardize analysis for large projects or core facilities, but are also useful for individual biologists who need to perform repetitive data processing. Some workflow systems, designed for beginners, offer a graphical user interface and have been very popular with biologists. In practice, these tools are infrequently used by more experienced bioinformaticians, who may require more flexibility or performance than afforded by the user interfaces, and seem to prefer developing workflows with scripting or command line tools. Here, we present a workflow system, the NextflowWorkbench, which was designed for both beginners and experts, and blends the distinction between user interface and scripting language. This system extends and reuses the popular Nextflow workflow description language and shares its advantages. In contrast to Nextflow, NextflowWorkbench offers an integrated development environment that helps complete beginners get started with workflow development. Auto-completion helps beginners who do not know the syntax of the Nextflow language. Reusable processes provide modular workflows. Programmers will benefit from unique interactive features that help users work more productively with docker containers. We illustrate this tool with a workflow to estimate RNA-Seq counts using Kallisto. The workflow can transparently run either on a laptop computer with docker or on a linux cluster. We found that beginners can be taught how to assemble this workflow in a two hours training session. In conclusion, the NextflowWorkbench simplifies the development of reproducible, implicitly parallel workflows. NextflowWorkbench is distributed under the Apache 2.0 license and available at http://workflow.campagnelab.org.