PHASING THE CHOLERA TOXIN ELECTRON DIFFRACTION DATA USING THE MAXIMUM ENTROPY-LIKELIHOOD METHOD WITH NON-CRYSTALLOGRAPHIC SYMMETRY IMPOSED IN THE MICE COMPUTER PROGRAM. W.N. Nicholson and C.J. Gilmore, Department of Chemistry, University of Glasgow, Glasgow G12 8QQ, Scotland U.K.

We have already reported our experience with applying the maximum entropy-likelihood method to phasing the two-dimensional projection data for cholera toxin (Gilmore & Nicholson (1995). Transactions American Crystallographic Association, 31 In press.). We have been working with a 2-d data set for which 56 unique image phases are available at 8.8Å resolution, and for which a further 1417 diffraction intensities extend to 4Å. The problem has been:

To phase the 4Å data from the 56 known phases.

To impose 5-fold non-crystallographic symmetry on the projection.

To impose envelope and solvent flattening constraints.

The maximum entropy-likelihood program (MICE) has been modified to carry out these steps. Using it we have shown that the likelihood criterion to is an accurate and reliable predictor of:

The effective number of atoms in the unit cell.

The centroid coordinates for the 5-fold non-crystallographic axes.

3. The envelope radius.

We are now extending the procedure to phase data derived from a series of electron diffraction patterns derived from a limited set of specimen tilts. The extension of two-dimension phase information into three is non-trivial especially when so little a priori phase information is available, and will be discussed in detail.

Research funded by a grant from the Human Frontier Science Program.