What is a biobox?
Biologists should not have to be experts in computer systems administration. Getting a bioinformatics tool to analyse your sequence or proteomics data should not require spending the afternoon debugging installation problems. Bioboxes solves this problem by creating pre-packaged versions of bioinformatics tools that are simple to install and simple to use. Each biobox is a Docker image containing a bioinformatics tool, with a standardised interface through which data and parameters are passed. This page explains what these terms mean.
Pre-packaged bioinformatics tools
Docker is a San Francisco company that has created a software platform with the same name. Docker allows the creation of "images", which can be thought of as a software box in which you can add everything you need to make a particular tool work. For example a Docker image of the velvet genome assembler includes Ubuntu Linux, extra libraries required for installation, velvet itself, and extra scripts to run velvet on input data. Packaging all these together into a Docker image makes it easier to share and obtain everything needed to run velvet. You don't need to go through the manual process of installing velvet yourself. This is key advantage that Docker provides: the process of giving your software, or using another persons software is now much simpler.
If you're familiar with virtual machines (VMs) you can think of Docker images as more lightweight versions of a VM. Starting a Docker image is much faster than a VM, as Docker images also the host machine resources rather than reserving a portion of host resources for their own use. If you have docker installed, here is an example of how simple it is to get the velvet Docker image:
docker pull bioboxes/velvet
This will fetch the velvet Docker image on to your computer and you can now begin using it as if it was installed, because in effect it is. You don't need to do any other set up or installation to start using the velvet Docker image in your work.
Simplifying the install process is only half of what a biobox is. The other half is making bioinformatics tools simple to use. This is done by standardising the tool interface.
a simplified user interface
Having a bioinformatics tool installed is no garantee of getting good results with your data. This requires experimenting with different combinations of paramaters and run time options. Again this means extra work for biologists and bioinformatics when most of the time they would like they are only interested in the results for further downstream analyses.
Bioboxes solve this usuability problem by making all tools of the same type have the same interface. For example all biobox genome assemblers accept input FASTQ the same way, and return assembled contigs the same way. This means that if you learn how to use one biobox assembler you can use all available biobox assemblers. This standardised interface makes it simple to swap one type of biobox for another in your workflow, a common situtation when algorithm improvement are published.
The biobox interface is a file passed to the biobox image specifying all the inputs needed to run. This file is called the 'biobox.yaml', an example for a genome assembler biobox is:
--- version: "0.9.0" arguments: - fastq: - id: "test_reads" type: "paired" value: "/bbx/input/reads.fq.gz"
This specifies the version of the file and the arguments. In this case the argument is a paired fastq file which will be mounted in the biobox at "/bbx/input/reads.fq.gz". YAML is used as a standard machine- and human-readable file format. The interface work where a user writes biobox.yaml file containing their input data and parameters, and then passes this to biobox. A biobox developer then expects the input data to be in this format, and so they are can build their tool around accepting this kind of input. Importantly this biobox.yaml file will work with all biobox assemblers, making it easy to swap one tool for another.