Evolutionary dynamics under a stability-constrained model
The space of possible proteins is vast. For all but the smallest proteins, the number of sequences exceeds the number of atoms in the observable universe. Evolution—through the forces of natural selection, drift, and mutation—samples this space, leading to proteins with diverse structures and functions. Evolutionary biologists interested in understanding the history of a protein must identify signals from patterns of substitutions and decipher their likely causes. However, the true evolutionary process is often unknown. Simulations of protein evolution allow us to investigate various emergent phenomena with complete knowledge of the generating parameters in hand. Additionally, using plausible simulating models, we can assess the accuracy of inference procedures which, by necessity, make simplifying assumptions about the process of sequence evolution. In this dissertation, I focus on stability constraints of proteins using a modelling framework grounded in the formalisms of thermodynamics and population genetics theory. In Chapter 2, I show that stability-constrained evolution recapitulates various patterns present in natural alignments. I demonstrate that epistasis due to stability leads to elevated substitution rates compared to site-independent evolution and discuss the underlying mechanisms causing this increase. Additionally, I investigate the accuracy of rate inference from commonly used inference models. While the amount of among-site rate variability is often underestimated, the inferred rates correlated with the most common rates across sites. In Chapter 3, I explore the dynamics of resident amino acid propensities and show that decreases in propensities can occur due to epistasis, challenging claims that such a trend must have adaptive origins. In Chapter 4, I conduct a literature review on nonadaptive phenomena that lead amino acid preferences to change over time. Finally, in Chapter 5, I investigate the evolutionary response to destabilizing substitutions across and within protein structures. I find that destabilizing substitutions at buried residues often require a longer time for the effects to be mitigated than destabilizations at exposed sites. I end the dissertation by discussing the implications of epistasis on protein evolution and future research directions.