4 September 2026
Predicting Evolution: a possible ARIA programme
As Programme Director Yannick Wurm explores a new programme, he shares an early draft of his thinking and invites you to provide feedback to help refine the concept.
Programme Director, Yannick Wurm, speaking at ARIA's 2025 Summit
Evolution is the process that makes the rest of biology explicable, yet evolutionary biology has typically proved far better at explaining the past than forecasting the future.
Yet there are situations where seeing evolution coming could make all the difference, when:
- Engineered organisms and cell lines drift from their designed function.
- Tumours escape the treatments we prescribe.
- Gene-edited cells delivered as therapy can fail to establish – or evolve beyond our control.
- Agricultural pests evolve resistance to pesticides faster than we can replace them.
- Pathogens and parasites evade drugs and vaccines.
- Invasive species adapt and establish in new territories.
What if we could transform evolutionary biology from a descriptive science, to a predictive tool?
We believe that three changes make a new approach possible:
- Sequencing has become cheap enough to track not just genomes but the molecular states that connect genotype to phenotype, across tens of thousands of populations.
- Robotics can automate evolution experiments at a scale no lab could previously run.
- And modern machine learning can find structure in data of this size and complexity.
This proposed programme would test whether, combined, these technological changes are enough to forecast which populations can evolve their way through new challenges, and which will be driven extinct.
In practice, we propose to achieve this through massively replicated experimental evolution in a model animal, such as the roundworm Caenorhabditis remanei, with open, blind competitions in which teams forecast which populations persist, which collapse, and the genetic changes behind those outcomes.
If successful, evolution becomes something we anticipate rather than explain in hindsight, wherever it determines whether a crop protection, a therapy or an engineered organism endures. Even partial success delivers something field-defining: a map of where prediction holds and where it fails.
We want to hear from individuals this programme would need – biologists who understand evolving populations, modellers and computer scientists who might forecast them, people who have run prediction competitions, and practitioners – in agriculture, medicine, conservation, biosecurity – who would use the answers.