PERFORMANCE ANALYSIS OF THE DOUBLE-ITERATED KALMAN FILTER FOR MOLECULAR-STRUCTURE ESTIMATION

A possible application of a novel double-iterated Kalman filter (DIKF) as an algorithm for molecular structure determination is investigated in this work. Unlike traditional optimization algorithms, the DIKF do es not exploit experimental nuclear magnetic resonance (NMR) constrain ts in a penalty function to be minimized but used them to filter the a tomic coordinates. Furthermore, it is a nonlinear Bayesian estimator a ble to handle the uncertainty in the experimental data and in the comp uted structures, represented as covariance matrices. The algorithm pre sented applies all constraints simultaneously, in contrast with DIKF a lgorithms for structure determination found in literature, which apply the constraints one at a time. The performances of both paradigms are tested and compared with those obtained by a commonly used optimizati on algorithm (based on the conjugate gradient method). Besides providi ng estimates of the conformational uncertainty directly in the final c ovariance matrix, DIKF algorithms appear to generate structures with a better stereochemistry and be able to work with realistically impreci se constraints, while time performances are strongly affected by the h eavy matricial calculations they require. (C) 1996 by John Wiley & Son s, Inc.