What can this program do for you?

This program is a virtual evolution machine that will simulate artificial selection on a population of protein models with identical amino acid sequences, although length variations are allowed. Typically these initial models have been built by the user on a one by one basis, for example using well known servers or programs like 3D-JIGSAW, SWISSMODEL, EsyPred3D, Modeller or Jackal, that give the user full atomic detail. In addition, fold recognition servers can be used (see for example metaserver), but backbones or side-chains may be needed to be constructed if they are to be used in this program. Scwrl and MaxSprout can be used for that.
Because selecting templates for modelling is not trivial often models using different templates (and very importantly, alternative alignments) need to be built. All these models are valuable information for in silico protein recombination. On the other hand, the more models you add to your initial population the more you will have to wait to see your results back.

If the program concludes successfully, the final population will contain protein models which have been optimized to approximate the native structure for this particular sequence. Usually final populations converge to unique conformations, but not always, because some times different conformations cannot be distinguished with our selection procedure.

The main strengths of the method are:
  • the ability to overcome alignment errors through recombination
  • templates and alignments are combined to yield optimized models
  • templates needn't to be selected before hand
    The main problems are:
  • This algorithm can be trapped in suboptimal populations and therefore it is advisable to replicate evolution experiments.
  • For the sake of speed, this method produces finals models that can have stereochemical problems and even broken bonds. So we strongly advice to minimize their geometric clashes afterwards (using CHARMM or gromacs for example).

    What are the parameters?

    population size must be larger than the number of proteins in your input file, but currently the program will only accept up to 200 members
    selected proportion proportion (0,1) of proteins that will be selected for the next generation after ranking them according to energies
    forced improvement required relative improvement of a newly created protein over the average population energy (0,0.1)
    don't put it too high or the program will never finish!
    last generation maximum number of generations that the simulation will run if convergence is not achieved
    mutation rate fraction of mutation over 100 (the alternative is recombination)

    home