Making model assumptions clear
This week I am in Columbus, Ohio, in a meeting organised by Sandy Anderson and Trevor Gram on tumour heterogeneity. It's always a pleasure to leave Tampa in the middle of the winter to fly north so I must confess it has been a productive week. Initially I was planning to just attend their meeting [link]. But unfortunately for the attendees I was asked to give a talk so I took this opportunity to talk about three different projects.
- The first one is a theoretical study of the role of hard edges and boundaries in tumours and how they change the evolutionary dynamics of the tumour. Artem Kaznatcheev has blogged extensively about this [link[link][link]. The gist is that edges select for different phenotypes than the core or the growing boundary.
- The second one is what is known as Team Science. Together with an oncologist, a pathologist, a radiation oncologist, a GU surgeon, an epidemiologist, experimental biologists and mathematical modellers, we are trying to change how prostate cancer patients with bone metastases are treated. The idea is that if we could measure heterogeneity in the patients tumour we could use a mathematical model to optimise their treatment. We have published the approach recently [link] so you can find more online.
- The third one is an agent-based model of prostate cancer in bone metastases. Is a more complex model than the one mentioned before. The idea is less to guide clinical treatment and more to try to understand biology. Knowing how the tumour grows in the bone will allow us to find how to stop it. The idea then is to try to capture all the key cell types and microenvironment features and see what tumour cells do with them. We have done a lot of new work with this model but a description of it can be found here [link].
It this last model that draw the most attention from the biologists. Many biologists do not mind how clever the mathematical model may look but do care about the assumptions we make. This is because it's easy to find papers in favour or against any assumption. So making these assumptions clear during the presentation is important. Without that many biologists will not trust us. I will argue though, that cancer biologists are working with models as well and they should do a better job of explaining the assumptions and limitations of their models. They do a good job of showing where these models excel. But although showing assumptions in an in vivo model is harder is also important.