About the efficiency of the early FOMs in ab initio protein phasing

All ab initio techniques for solving protein crystal structures use multisolution approaches. Several figures of merit that are found in the literature are efficient in the last steps of the phasing process, when some trials converge to the correct solution with a relatively small average phase error. Early figures of merit are much more critical; they should be able to recognize useful trials when the phase error is still large, and their efficiency determines the efficiency of the program. In the present work, a wide variety of known figures of merit at atomic and quasi-atomic ( similar to 1.4 Angstrom) resolution have been tested; new figures have also been devised and tested. Their application to a large set of test structures allows the study of their properties at different data resolutions and the selection of the most efficient figures within the SIR2003-N framework.