Firstly, vast amounts of data have become available due to maturation of high throughput technologies such as the next generation sequencing (NGS) and massively parallel phenotyping assays. These technologies enable tracking the evolution of real or laboratory populations in (near) real time. Secondly, increasingly powerful theoretical methods to analyse and make sense of these data have been developed. This progress underpins our research that focuses on the main themes of
Rapid evolution can underlie the response of populations and communities to changing environments, including the therapy response of infectious diseases and cancers. A quantitative understanding of rapid evolution can therefore be used to design better therapeutic protocols. This requires an integration of population genetics, ecology and predictive modelling, and can be informed by data generated through in vitro evolutionary experiments. Microbial models are particularly suited to experimental purposes, and can be used to produce both general insights into the predictability of rapid evolution and insights directly relevant to combatting the global antimicrobial resistance crisis. Several of our ongoing projects, pursued together with our collaborators, seek these insights by combining experimental evolution using microbial systems with data science. For instance, we will find out how much information the drug exposure history of a population contains about its future phenotypic and genomic status.
Some evolutionary changes are straight forward to predict: a pathogen population either is or becomes resistant to a given antibiotic, or goes extinct. Longer selection sequences, on the other hand, are more unpredictable. Can the future evolutionary path of a population be predicted based on its current state, together with previous experiments and other observed data? Moreover, if we know the most probable evolutionary trajectory, could we change it by intervening evolution (e.g., by modifying a treatment), given that the current trajectory is predicted to lead to an undesirable outcome. To address such questions we rely on mathematical modelling combined with a relevant biological background. We develop, analyse and simulate appropriate temporal or spatiotemporal stochastic models accompanied by effective use of control theory and machine learning methods.
Any therapy or intervention against a fast-evolving pathogen or cancer is an attempt to control its future population. What building blocks should optimal treatment strategies in cancer or bacterial infection contain? Most cancer studies and mathematical models focus on overall cancer cell population growth, which can be realised through various combinations or cell reproduction and cell death rates, reaching from fast to slow cell turnover. Several recent studies have identified cell population turnover as a key factor in cancer evolution and tumour formation. We are taking a deeper look into the underlying birth and death processes, utilising mathematical modelling as a tool to explore the effectiveness of chemotherapy under different turnover scenarios.
Collaborating groups: Tero Aittokallio, Trevor Graham, Teppo Hiltunen, Gianni Liti, Michael Lässig, Danesh Moradigaravand, Leo Parts, Jonas Warringer.
For more information about related completed projects see: