SOFTWARE FOR VERY LOW-RESOLUTION AB-INITIO PHASING

N.L.Lunina

Institute of Mathematical Problems of Biology, Russian Academy of Sciences, 142292 Pushchino, Moscow Region, Russia; e-mail: lunina@impb.serpukhov.su

Kewords: ab-initio phasing, low-resolution images, macromolecules

The solution of the phasing problem at very low-resolution requires non-standard approaches and software. The Few Atoms Model (FAM) approach, suggested recently[1-3], is one of such methods.

The first stage of the approach consists in the generation of a large number of FAMs (every one of which is formed by a relatively small number of gaussian spheres), calculation of the corresponding structure factors (s.f.) values and selection of the variants resulting in a good correlation of the calculated s.f. magnitudes with the observed ones. The proposed computer program allows to use at this stage different prior probability distributions, to apply additional stereochemical restrictions and to apply multiple criteria to the choice of variants.

As it was shown in test cases, the best values of formal agreement criteria do not necessarily correspond to the true solution at very low resolution. Therefore, the next stage consists in the clusterisation of selected phase sets into groups of neighbouring ones and the calculation of cluster centroid phases and figures of merit. To solve the problem of origin and enantiomer choice, the special procedure of phase sets alignment [4] is applied here. The programs developed allow to perform the cluster separation in interactive graphic mode. The output of this stage is a small number of s.f.-sets corresponding to alternative solutions of the phase problem.

The most delicate problem is the choice of the most likely solutions from the set of isolated clusters. To solve this problem, different approaches were introduced into the software. The first one is based on the generalized likelihood principle [3] and consists in the generation of atomic models inside different envelopes and the comparison of search criteria histograms. The second one is connectivity analysis, i.e. the connected regions in different electron density maps are found and preference is given to maps consisting of a reasonable number of globular regions.

One of the problems where the FAM approach may be used is the application of the molecular replacement method at very low resolution. Special programs were developed for the search of the best fitting of EM-model to FAM-phased syntheses.

The developed software was used in the testing of methods and applied for very low-resolution phasing of ribosomal particles [5].

The work was supported by RFBR- 97-04-48319 grant.

1.V.Y.Lunin, N.L.Lunina, T.E.Petrova, E.A.Vernoslova, A.G.Urzhumtsev, A.D.Podjarny: Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography, 30 (1994), 37-44.
2.V.Y.Lunin, N.L.Lunina, T.E.Petrova, E.A.Vernoslova, A.G.Urzhumtsev, A.D.Podjarny: Acta Cryst, D51 (1995), 896-903
3.V.Y.Lunin, N.L.Lunina, T.E.Petrova, A.G.Urzhumtsev & A.D.Podjarny: Acta Cryst.(1998) (to appear). 4.V.Y.Lunin, N.L.Lunina: Acta Cryst, A52 (1996), 365-368.
5.V.Y.Lunin, N.L.Lunina, T.E.Petrova, A.G.Urzhumtsev, A.D.Podjarny: ECM-18 presentation.