MONTE CARLO METHODS IN PROTEIN RECURRENT DESIGN
Alina BUŢU
*, Marian BUŢU
**
* National Institute of Research and Development for Biological Sciences, Bucharest
** GRS COMMUNICATION, Bucharest
Abstract. In this work we present a novel approach to the protein design problem, our principal contribution being the ability to specify the degree of accuracy desired in the solution set. We achieve this by replacing the exhaustive search steps of the original recursive design algorithm with Monte Carlo sampling of conformation space. A modification to the recursive design of proteins in the 2D-HP model is made, this algorithm finds supersets of the solution set, and by varying parameters can minimize the size of this superset to any desired accuracy.