Home > Virtual machines enhance the peer-review process
Virtual machines enhance the peer-review process
13 Dec 2016 - 09:15
Research that relies heavily on bespoke code and software can pose a problem for journal referees: how do they test the methodology without running the code? A range of challenges often prohibits such rigour in the peer-review process. To overcome this problem, UCT eResearch collaborated with Dr Musa Mhlanga of the Biomedical Translational Research Initiative (BTRI) to use virtual-machine technology to allow referees access to both code and data.
“Too frequently though, our guidelines notwithstanding, papers reporting a computational method are initially submitted without either [software or code]. Separately, referees quite frequently have trouble running the provided software, which they may report to us during the review process or at the end of it. Both of these problems can lead to delays, or even the rejection of a paper.”
It was with these challenges for the peer-review process and publication in mind that a team of researchers in the BTRI, led by Dr Musa Mhlanga, collaborated with UCT eResearch to deploy a virtual machine, to allow referees access to both the computational code and the data central to their methodology.
“In a way, a virtual machine is a precursor to cloud computing,” says Dr Jason van Rooyen, eResearch analyst. “Instead of having individual computers in a laboratory with all the required software and data loaded, we at UCT eResearch run a server in our facilities that offers remote access to the software and data to anyone who needs it, be it collaborators or – in this case – referees.”
The paper in question was submitted to the journal Cell as a resource article, the purpose of which is to highlight significant technical advances.
In their research, Mhlanga and his team developed a completely novel approach to the measurement of spatial patterns of RNA on a subcellular level.
“RNA (ribonucleic acid) is known as the ‘origin of life’ molecule. It is believed to predate DNA and protein,” says Mhlanga. Part of his research is to understand the spatial distribution of RNA in cells.
“This spatial distribution – while very poorly understood – is important for several cellular processes, including development, immune response, and learning and memory,” he says.
So the group embarked on a systems biology approach, to try to understand how RNA is spatially localised.
“To do this we developed a mathematical approach, combined with imaging and microscopy, which led to fundamental insights into how cellular organelles influence the spatial distribution of a broad class of RNAs,” he explains. Because of the novel tools used to gain what Mhlanga calls “a provocative new insight”, it was critical the referees could access the data and run the code.
One challenge is the requirement for anonymity of referees, which for Mhlanga is why the collaboration with eResearch was so critical. “We gave anonymous accounts for access to the virtual machine to Cell, which meant referees could assess the code and the data, comfortable that their anonymity would be maintained,” he explains.
“This step towards allowing referees easy access to review code and to run software has the potential to add a new depth to the peer-review process, and help ensure more reproducible science,” says van Rooyen.