T-World 7: Concluding remarks

That’s it, dear readers.

I can’t thank enough the people who contributed to this project and helped shape T-World into its final form. It was wonderful to work on this with a range of very different people, addressing different aspects of the model development and presentation, from support in formulating details of calcium handling, conceptualizing new components, to demonstrating applicability to interesting and relevant tasks. I’ve been going through memories of different stages of the project, and the project wouldn’t be the same without you all! I am also very grateful to the Wellcome Trust, whose fellowship gave me the freedom to focus properly on the later stages of development and bring the work to completion.

At the same time, this project has made me think a bit about how we support the development of substantial computational methodologies. My experience with T-World was ultimately a positive one, but it also highlighted how difficult this kind of work can be to carry out within the current system. Projects of this sort often require long development cycles, repeated redesign, extensive validation, and a great deal of work that does not fit neatly into short-term application-driven narratives. I managed to do this because I had postdoc supervisors who allowed this in the first phases. I worked on this a lot in the small remnants of free time I had after our kids were born, and then I had a fellowship that gave me enough freedom to finish this. Also, I did not mind assembling a custom workstation using my personal money, and burning thousands of pounds in energy needed for the simulations. To see strong computational work of this sort in the future, I think it would be beneficial to support this kind of methodology development more explicitly.

First, it would help to have funding schemes or fellowships specifically aimed at the development and validation of new computational methods over longer timescales. Not every project of this type should have to be packaged primarily as an immediate application story (which usually does not incentivize generality). Development and validation are substantial scientific goals in their own right, provided one can clearly articulate why the method matters and what kinds of downstream questions it could enable.

Second, I think the community should place greater value on the careful characterization of the tools we build, and of the underlying real systems. Applied studies are essential, and of course we all want to see research translated into biological or clinical impact. But progress also depends on work that is less flashy and more foundational: understanding model behaviour, identifying causes of failure, refining assumptions, and establishing physiological relevance. Achieving this often warrants carrying out experiments on the model system, that guide further development. Those efforts may be less glamorous, but they are often what determines whether a methodology becomes useful and innovative in the long run.

Third, for some kinds of computational work, access to strong local computing resources (workstation-type) can make an enormous difference. High-performance computing is indispensable for many tasks, but it is not a substitute for numerous tasks such as model development, where rapid cycles of design, testing, visualisation, and adjustment are essential. When properly justified, support for that kind of infrastructure can significantly improve both the pace and the quality of methodological research.

T-World did make it through, and I am very glad it did. But I also think the experience points to a broader opportunity: if we want more robust, ambitious, and well-validated computational methodologies, then we should make it easier for researchers to build them in a way that is supported by the system. I hope this project provides not only a useful model, but also a case for taking that challenge seriously. OK, that’s it – we will see where things go next, as this is a period of uncertainty – for now, thanks for reading!