ViPAR - Virtual Pooling and Analysis of Research data
Research studies exploring the determinants of disease require sufficient statistical power to detect meaningful effects. Sample size is often increased through centralised pooling of disparately located datasets though ethical, privacy, and data ownership issues can often hamper this process. Methods that facilitate the sharing of research data that are sympathetic with these issues and allow flexible and detailed statistical analyses are therefore in critical need.
Methods and Results
We have created a software platform for the Virtual Pooling and Analysis of Research data (ViPAR), which implements database federation techniques to provide researchers with a web-based platform to analyse datasets housed in disparate locations without the need for permanent central storage. Remote sites manage their own harmonised research datasets in a database hosted at their site, while a central server hosts the data federation component and a secure analysis portal. When an analysis is initiated, requested data are retrieved from each remote site and temporarily pooled into computer memory at the central site. The data are then analysed by statistical software, and on completion, the results of the analysis are returned to the user and the virtually pooled data are removed from memory.
ViPAR is a secure, flexible, and powerful analysis platform built on open source technology that is currently in use by two large international consortia, and is made publicly available at http://bioinformatics.childhealthresearch.org.au/software/vipar/.
The easiest way to get started with ViPAR, is to read the Quick Start guide here: ViPARQuickStart_1.0.0.pdf
To get started with ViPAR, please download one of the following pre-built images (guest VM is a 64bit Ubuntu 12.04 system). Please note, these two VM images contain both the VMS/VWAP master site components, and 3 test sites (LVDs) installed and pre-configured.
- The ViPAR Manual: ViPARManual_1.0.0.pdf
- VirtualBox (4.3+) compatible VM image (1.4Gb in size): VIPAR-VBOX-1.0.0.zip OR
- VMWare player (6+) compatible VM image (1.5Gb in size): VIPAR-VMWARE-1.0.0.zip
If you would like to test ViPAR with separate VMs (or physical machines) ie one server running the VMS&VWAP components and a separate one acting as a remote LVD, you can also download a separate prebuilt VirtualBox LVD image following. This option does require a more complicated network setup to get working, please read the Manual and visit the Google Groups (Help and Support) below for more information.
- LVD-only VirtualBox (4.3+) compatible VM image (1.4Gb in size): VIPAR_LVDONLY_VBOX-1.0.0.zip
The source code for ViPAR is made available freely at our Git repository: https://gitlab.com/kim.carter/ViPAR. The easiest way to test and use ViPAR is to start with one of the pre-built images - as we've done all the hard work installing and configuring all of the necessary operating system and software packages to make it just work. If you would like to create your own ViPAR installation, please visit the Git repository and see the Manual for detailed dependancies. We also strong encourage you to join our Google Groups list.
Help and Support
If you run into any problems with ViPAR, we encourage you to have a detailed read of the Manual and to join our online support forum at: https://groups.google.com/forum/#!forum/vipar. When we come across a potential issue / solution that might be of relevance to more than one person, we will be posting answers/solutions. We also encourage you to post your questions / feature request, and importantly any bugs/errors you might find.
In Windows 8.1 we are aware of an issue with running 64 bit guest virtual machines (such as ViPAR). Please see our Google groups forum for a solution