---

In mathematical oncology, we are very familiar with the idea that it is valuable to attack challenging biological problems using a combination of perspectives. Clinical trials give us data about what actually works for real patients, but there are immense practical and ethical restrictions on what questions we can use them to ask. Mathematical models are fast and easy to interrogate, but may miss important dynamics that would occur in a patient. These two study systems represent opposite ends of a continuum (see Figure 1), in which we can generally gain speed, control and generalizability by sacrificing “naturalness” of the dynamics in our system. Each point along this continuum allows us to ask questions that aren’t possible at other points. By using this full spectrum of study systems, we can build a more accurate mental model of how we expect cancer to evolve.

Figure 1: The spectrum of cancer research systems. Note that the boundaries between many of the categories noted here are fuzzy, and that many additional categories could be slotted in. For example, “laboratory experiments” could be broken up into in vitro experiments (e.g. cells in a dish) and in vivo experiments (e.g. mouse models); the former would be to the left of the latter on this spectrum. Figure adapted from¹.

One point along this spectrum that may not be familiar to everyone is digital evolution. In this paradigm, we create complex evolving populations in a computer by encoding a set of ground rules sufficient to produce evolution by natural selection. From there, we treat our digital system much like a model system in the wet lab, conducting in silico experiments. As a result, we have a system that displays richer evolutionary dynamics than a typical simulation, but that enables us to conduct experiments that wouldn’t be possible in the laboratory.

Digital evolution has a long history of success in evolutionary biology research, but has less frequently been applied to scenarios with interesting ecological dynamics. In our recent review paper¹, we 1) review ecologically-relevant digital evolution research to date, and 2) explore the potential of digital evolution as a tool for studying eco-evolutionary dynamics. While the paper is focused on more traditional ecology, almost all of the points that we make in it apply to cancer research as well – particularly in light of the increasingly evident importance of ecology for predicting cancer evolution.

There is one important caveat to be aware of: digital evolution is intentionally unlike any specific biological system. This trait can be a strength, in that results observed in both digital evolution are a biological model system are likely to be highly generalizable. However, it can be a weakness when the goal is to draw inferences about a specific biological system. Some of the questions we find ourselves asking in mathematical oncology relate to a specific cancer, and so may not be well suited to digital evolution. However, others are very general eco-evolutionary theory questions that happen to arise in the context of cancer. For these, digital evolution is a compelling approach. It allows us to relax the rigid assumptions of our more-targeted simulations and ensure that our theory is robust before moving to more costly and logistically challenging study systems.

References

1. Dolson, E., & Ofria, C. (2021). Digital Evolution for Ecology Research: A Review. Frontiers in Ecology and Evolution, 9, 852. https://doi.org/10/gnp8nc