LMP Seminar: How does the structure of a genotype-phenotype map shape variation and evolutionary processes?

Evolutionary processes are relevant in many areas of the life sciences, and since these processes are fundamentally quantitative and stochastic, mathematical and computational models are required to capture their dynamics. Such models need two central elements: variation through random mutations and natural selection. To connect variation and selection in a single model, we need to translate random mutations on the level of genotypes to their selectively relevant phenotypic consequences. This connection is given by genotype-phenotype (GP) maps, datasets translating any possible genotype to its phenotype. GP maps are large (with 4^L genotypes of length L) and thus well-studied GP map examples are typically computational models of molecular phenotypes (for example, RNA secondary structures). However, further computational and empirical GP maps for phenotypes on the molecular scale and beyond are also becoming available. Therefore, it is important to take a systematic approach that can be applied to many GP maps: This approach should identify coarse-grained features that characterise a GP map, and investigate, how they shape evolutionary processes on this map. In this talk, I will illustrate this with two examples [1, 2]: (1) How does one specific feature called “genetic correlations”, which had been reported in several GP maps, shape evolutionary processes? (2) What are the topographies of GP-map-based fitness landscapes (in the baseline case of random phenotype-fitness assignments); can we break these down to specific GP map features?

[1] N. S. Martin, S. Schaper, C.Q. Camargo, A. A. Louis. Mol. Biol. Evol. 41(6):msae085 (2024)
[2] M. Srivastava, A.A. Louis, N.S. Martin. bioRxiv (2025)