Genetic algorithms representation scheme

In the past, different methods of sampling of protein model conformational space have been employed with various degrees of success. Traditional molecular dynamics can be used only in the case of continuous models. Other sampling schemes, including a variety of Monte Carlo methods, genetic algorithms, and combinations of these methods, could be applied to continuous as well as to the discrete (including lattice representation) models. 

In order to tackle the sample impoverishment problem, Chatzi and Smyth (2013) proposed an enhancement of the PF, termed the particle filter with mutation (MPF), which incorporates a mutation operator in the resampling process. Mutation is a process typically used in genetic algorithms (GAs) where it serves as a means of maintaining diversity among the members of a population. Within the framework of GAs, mutation is typically enforced under two regimes, the creep mutation and the jump mutation. Jump mutations involve random modifications in the binary encoding of the system s variables, whereas creep mutation takes place in the real number representation of the variables (i.e., the phenotype). The operation implemented in the MPA scheme resembles the creep mutation process. 

---