The first unit of the lecture dealt with the evolutionary problem of phenotypic innovation, i.e. trying to understand why the consequences of random genetic mutations are not random. One answer is that the modification of a phenotype is made possible by development, i.e. it depends on the mapping from genotype to phenotype. A computational model of this mapping, namely the folding of RNA sequences (" genotype ") into minimum free energy RNA secondary structures (" phenotype "), reveals some interesting properties.
Neutral networks
The set of sequences folding into the same shape is mutationally connected and extends a network across sequence space. Neutral networks enable populations to spread across genotypic space without losing their current phenotype, while allowing the exploration of new phenotypes accessible by mutation at the network's periphery. Neutral networks illustrate how robustness enables change. The impact of neutral networks on evolutionary dynamics appears in computer experiments as evolutionary trajectories with sudden transitions.
Neighborhood
Contiguity between neutral networks mathematically defines a notion of phenotypic neighborhood that is not based on similarity and is sufficient to define continuity and discontinuity in evolutionary trajectories.
Shape space coverage
All frequently occurring forms occur within a relatively small neighborhood of a random sequence. This is reminiscent of the asymptotic equipartition property, or notion of typical set, in Shannon's source coding theorem.
Plasticity reflects variability
By generalizing the definition of phenotype to include the set of alternative structures (" excited states ") in a thermal energy band from the minimum free energy (" ground state "), we observe that genetic variability (the set of ground states realized in the mutational neighborhood of a given sequence) correlates with plasticity (the set of excited states achievable by that same sequence).
