Role of random genetic drift in the evolution of interactive systems

Through comparative studies of DNA sequences it has become possible to test the neutral and the selection theories of molecular evolution. The separate estimation of the numbers of synonymous and non-synonymous substitutions is one of the most powerful tools for detecting selection. The patterns on the average and variance of these two types of substitutions of mammalian genes turned out to be in accord with the slightly deleterious or the nearly neutral mutation theory for non-synonymous changes. Interactive systems at the amino acid level were suggested to be responsible for such nearly neutral or very weak selection. An attractive model is the NK model of Kauffman, which assumes that each amino acid makes a fitness contribution that depends upon the amino acid and upon K other amino acids among the N that make the protein. It is known that the fitness landscape is very rugged for K > or = 2. Population genetic analysis of this model suggest that protein evolution obeys the nearly neutral theory and that random genetic drift is important. In other words, evolution becomes rapid in small populations because proportion of near-neutrality increases among new mutations, and proteins as interactive systems evolve by shifting through random genetic drift on the multipeaked fitness landscape.