|Degrees:||B.Sc. (Université à Montréal), M.Sc. (Université de Montréal), Ph.D. (Université de Montréal)|
Hypermutability of transitions within the CpG context due to spontaneous deamination of 5-methylcytosines is a process known to affect vertebrates, but not well characterized at that evolutionary scale. We take advantage of the availability of several vertebrate genomes to characterize the heterogeneity of the mutation rates of the CpG context for hundreds of proteins in several vertebrate groups using an inference-based simulation approach that we recently developed for phylogenetic analysis (Conditional Approximate Bayesian Computation). This approach enables to estimate CpG hypermutability, while accounting for amino acid preferences related to the primary structure of the proteins studied. We also use comparative phylogenetic methods to test if genomic aspects specific could explain variation of CpG hypermutability rates along the vertebrate tree, such as the genomic proximity of the studied genes to CpG islands.
Although simulations can be an integral part of molecular phylogenetics, their use remains under-utilized compared to what simulation studies can bring to the production of scientific knowledge. Simulations are useful for validating newly developed phylogenetic models by quantifying model error. They can also be used to assess our ability to detect the presence of a particular evolutionary mechanism (e.g., mutation, selection, drift) such as the presence of Darwinian selection by evaluating how much signal is available under “realistic” conditions. Some simulation algorithms (e.g., jump-chain methodology) are less computationally expensive and allow to simulate under richer phylogenetic models than the ones available for inferences. For example, it is easier to simulate under a model that incorporates aspects of interdependencies between sites of a sequence alignment (e.g., protein tertiary structure, CpG hypermutability) than implement those interdependencies within an exact Bayesian modeling framework using Markov chain Monte Carlo techniques. Also, our ability to simulate under richer models can potentially allow us to demonstrate that the presence of confounding effects, like CpG hypermutability, could invalidate our ability to detect specific evolutionary mechanisms, such as detection of Darwinian selection.
Laurin-Lemay, S, Rodrigue, N, Lartillot, N and Philippe, H, 2018, Conditional Approximate Bayesian Computation: A New Approach for Across-Site Dependency in High-Dimensional Mutation–Selection Models. Molecular Biology and Evolution, vol. 35, no 11, p. 2819‑34. doi:10.1093/molbev/msy173.
Laurin-Lemay, S, Philippe, H and Rodrigue, N, 2018, Multiple Factors Confounding Phylogenetic Detection of Selection on Codon Usage. Molecular Biology and Evolution, vol. 35, no 6, p. 1463‑72. doi:10.1093/molbev/msy047.
Laurin-Lemay, S, Angers, B, Benrey, B and Brodeur, J, 2013, Inconsistent genetic structure among members of a multitrophic system. Bulletin of Entomological Research, vol. 103, no 2,p. 182‑92. doi:10.1017/S000748531200051X.
Laurin-Lemay, S, Brinkmann, H and Philippe, H, 2012, Origin of land plants revisited in the light of sequence contamination and missing data. Current Biology, vol. 22, no 15, p. R593‑94. doi:10.1016/j.cub.2012.06.013.