Hierarchical Minimization with Distance and Angle Constraints

Proceedings Of The Sixth International Conference On Intelligent Systems For Molecular Biology

Volume

Proceedings Of The Sixth International Conference On Intelligent Systems For Molecular Biology

Contents

The incorporation of experimentally-determined constraints into structure-prediction methods based on energy minimization leads to both improved selectivity with empirical potential functions and structure determination with far fewer constraints than are required for distance-geometry calculations. Some methods will be described for using both distance and angle constraints with the hierarchical minimization algorithm. The simulation is based on a combination of Monte Carlo Simulated Annealing and Genetic Algorithm techniques which are integrated into a single framework. The selection cycle of the genetic algorithm is carried out at the same temperature as the mutations, or alternatively the crossover cycle can be considered as a type of Monte Carlo trial move, such that each temperature annealing step corresponds to a new generation. The sequence is divided up into segments, and the mutation step consists of replacing an entire segment with a choice from a pre- selected list. This list is in turn constructed from a list of smaller segments, and the number of overall conformations can thus be pruned at each level of selection. Results will be shown for test cases using a small number of flexible distance constraints used as an additional term in the potential, and for restrictions placed on backbone dihedral angles used as an additional screening criterion for constructing trial moves.