Prof. Dr. E. Bornberg-Bauer
Principles of molecular evolution and genotypic variation
Simplified biophysical models of biopolymers (such as lattice proteins or the RNA secondary structure model), are a computationally tractable method to
investigate principles of molecular evolution. Although representing a highly idealised world, many important features of real-world biopolymers, such as databases
searches with profiles or consensus sequences, organisation of families as networks (dense clusters) in sequence space, structural robustness against mutations, the
dominance of a few folds or the domain-like arrangement of families, can be explained in terms of neutral evolution, punctuated equilibrium and recombination
events. As
a key feature we find that genotypes (sequences) which encode a structure uniquely arrange in neutral networks, i.e. all genotypes are connected by paths comprising
minimal mutational steps such as the hamming distance of 1 (point mutation). These networks arrange around a prototype sequence, which is the sequence which
allows for the largest number of neutral mutations within the neutral net and for which the ground state structure is not only unique but also maximally stable. With
increasing mutational distance from the prototype sequence, stability decreases and degeneracy becomes more likely. Most strikingly, this gradient of stability - which
we term superfunnel - is not confined to the neutral net but spans all geneotype space such that at any given point in genotype space it requires only a few
mutations to find a "direction" in which evolution must proceed in order to select for any structure related fitness criterion.
We have exploited this feature to design an
algorithm which, in combination with a novel constraint programming algorithm (designed by Backofen and Will, Univ. of Jena), enables the fast and reliable location of
unique and stable structures from starting anywhere in sequence space. Here we use unconstrained structure representations and the optimal solution is guaranteed
to be found within less than a second for sequences of length up to 30 and a four-letter alphabet. We find that many claims about the validity of certain models are wrong
(e.g. the statement that no unique structures exist in 3D for the HP model) and that such an evolutionary optimisation strategy should be applicable for the rational
design of biopolymers with certain features.
An important side aspect are investigations
on the choice of the algorithm which appears to be of minor importance compared to the choice of the biophysical model as far as evolutionary issues and
principles are concerned.
Our results have already successfully guided experimentalists for the rational design of proteins with desired functions (bistable structures) and further
experimental confirmation for the design of proteins has been suggested.
Another important point in our research is to explain how genotypic diversity is maintained during evolution although, intuitively, fitness optimisation should
narrow down variation. We do this by investigating the impact of multiple fitness criteria on biopolymer evolution and looking how rapid adaptation is facilitated when
evolution generates only marginally stable structures.
Beteiligte Wissenschaftler:
Kooperationen:
Veröffentlichungen:
|
