Distance geometry optimization for protein structures

We study the performance of the dgsol code for the solution of distance geo metry problems with lower and upper bounds on distance constraints. The dgs ol code uses only a sparse set of distance constraints, while other algorit hms tend to work with a dense set of constraints either by imposing additio nal bounds or by deducing bounds from the given bounds. Our computational r esults show that protein structures can be determined by solving a distance geometry problem with dgsol and that the approach based on dgsol is signif icantly more reliable and efficient than multi-starts with an optimization code.