Models in molecular evolution: the case of toyLIFE

Mención Internacional en el título de doctorThis thesis set out to contribute to the growing body of knowledge pertaining models of the genotype-phenotype map. In the process, we proposed and studied a new computational model, toyLIFE, and a new metaphor for molecular evolution —adaptive multiscapes. We also studied functional promiscuity and the evolutionary dynamics of shifting environments. The first result of this thesis was the definition of toyLIFE, a simplified model of cellular biology that incorporated toy versions of genes, proteins and regulation as well as metabolic laws. Molecules in toyLIFE interact between each other following the laws of the HP protein folding model, which endows toyLIFE with a simplified chemistry. From these laws, we saw how something reminiscent of cell-like behavior emerged, with complex regulatory and metabolic networks that grew in complexity as the genome increased. toyLIFE is, to our knowledge, the first multi-level model of the genotype- phenotype map, compared to previous models studied in the literature, such as RNA, proteins, gene regulatory networks (GRNs) or metabolic networks. All of these models either disregarded cellular context when assigning phenotype and function (RNA and proteins) or omitted genome dynamics, by defining their genotypes from high-level abstractions (GRNs and metabolic networks). toyLIFE shares the same features exhibited by all genotype-phenotype maps studied so far. There is strong degeneracy in the map, with many genotypes mapping into the same phenotype. This degeneracy translates into the existence of neutral networks, that span genotype space as soon as the genotype contains more than two genes. There is also a strong asymmetry in the size distribution of phenotypes: most phenotypes were rare, while a few of them covered most genotypes. Moreover, most common phenotypes are easily accessed from each other. We also studied the prevalence of functional promiscuity (the ability to perform more than one function) in computational models of the genotypephenotype map. In particular, we studied RNA, Boolean GRNs and toy- LIFE. Our results suggest that promiscuity is the norm, rather than the exception. These results prompt us to rethink our understanding of biology as a neatly functioning machine. One of the most interesting results of this thesis came from studying the evolutionary dynamics of shifting environments in populations showing functional promiscuity: our results show that there is an optimal frequency of change that minimizes the time to extinction of the population. Finally, we presented a new metaphor for molecular evolution: adaptive multiscapes. This framework intends to update the fitness landscape metaphor proposed by Sewall Wright in the 1930s. Adaptive multiscapes include many features that we have learned from computational studies of the genotype-phenotype map, and that have been discussed throughout the thesis. The existence of neutral networks, the asymmetry in phenotype sizes -and the concomitant asymmetry in phenotype accessibility- and the presence of functional promiscuity all alter the original fitness landscape picture.Programa Oficial de Doctorado en Ingeniería MatemáticaPresidente: Joshua Levy Payne.- Secretario: Saúl Arés García.- Vocal: Jacobo Aguirre Arauj