Marco Tripodi is a Principal Investigator at the Medical Research Council in Cambridge, UK, heading a research group in Neurobiology.
In a few words, what is your field of research and what are you working on?
We try to understand how neural circuits control movement and in particular directional movement towards a particular target. For example, if you see on a table a cup of tea that you want to drink you need to coordinate the position of this particular target with the correct motor-routine to reach that target. This requires to combine the awareness of the location with the motor control to reach that particular point in space.
But you don’t work with tea cups and humans? Can you explain a typical experiment and what kind of data do you normally collect for your research?
Yes, we normally work with mice and we use their head movement as a proxy for directional motion – so if the mouse reorients its head to the target that’s interested in, and realigns its body, in terms of motor control that corresponds to the movement that we perform when we want to reach for a cup of tea. So, that’s what we do with the mice: we give them a moving target, train them to chase it and if they reach it they get a reward. As they perform the chasing behaviour, we collect data on the position of the head and the movement in 3D. And then we record the underlying brain activity: to that end, we implant electrodes in the particular brain regions we are interested in and listen to the neurons. This way we aim to correlate the neuronal activity we record with the activity the animal is performing.
When publishing your results, are you sharing the methods or raw data in any form – is there a standard format for sharing data in your specific research community?
That’s a tricky one, I am very happy to share my raw data and some research groups share their raw data on their own website – but there is no standard way to do that. No journal currently requires to provide the raw data. Some journals have a sort of repository that allow other scientists to use the data, but most journals don’t. Equally there is no standard platform to share methods; every neural recording system we use comes with its own proprietary software for normalisation and analysis. The other way to do that is scripting. In all the in vivo experiments we are recording neural activity and in principle the raw data from all these recording systems could be transformed with scripts (in Python or Matlab) to make them comparable. We use GitHub, (an open online repository to share code), to make our scripts available to others; and some other groups in our research community do the same, but it’s still a niche thing.
So, the code written for analysis in another research group is applicable to analyse your data – is the raw data people obtain in in-vivo experiments similar enough?
Yes, if we wanted to standardize, we would for example agree to use Matlab or Python as a platform, we would just have to add a little script to convert the data from the proprietary file type to a standard file type and everybody could look at it. That would be very simple. You would need to provide the raw data file of the neural recording, a description of what file contains, basically what is in each column, and a script, which processes the data. Also, the sizes of the raw data files isn’t large: although we are recording over time, sometimes for hours, it is all numeric or words, so in practice we are talking about something in the range of a few megabytes.
Do you think you and others in your field would profit from a unified way of sharing data?
Yes, I think that would be fantastic if together with the paper, you could look at the data. It would be beneficial for standardising procedures and data collection routines, for the colleges, journal editors, students – everybody would benefit from it. I also don’t think there would hesitation to share data if there was a requirement from the journal – I mean the data are there, scientists are happy with what they did and proud of their research, so I don’t think there should be any problem.
Do you think a journal like Scientific Data can help in this process?
I think yes, it should go together with the journal because you need to create a reference set that then makes it easier to find the data that goes together with the paper. Indeed, it would be very straightforward in our field; you would just have to create a database, or even just a link to a folder, where you have the raw data together with the conversion-scripts (to read it into Matlab or whatever) and other scripts that process the data further. That would basically be everything you need.
Interview by Sylvia Tippmann, a freelance journalist based in London, UK