Efficient conformational space sampling for nucleosides using internal coordinate Monte-Carlo simulations and a modified furanose description

Internal coordinates can be very helpful in modeling large biomacromolecules because freezing stiffer degrees of freedom, such as bond lengths, strongly reduces the number of variables describing the system. This, however, leads to difficulties in treating flexible rings such as the furanose sugars of nucleic acids or the proline residues of proteins, for which internal coordinates are an overcomplete description. We present here a new, internal coordinate furanose model based on the pseudorotational variables phase and amplitude which avoids having to solve a ring closure problem. The choice of a two- rather than a four-variable description is justified by a detailed analysis of molecular dynamic simulations. The efficiency and accuracy of the method are also demonstrated using extensive Monte Carlo simulations. This method of ring treatment is fast and well adapted to macromolecular simulations. (C) 1995 by John Wiley & Sons, Inc.