TREATMENT OF MAD DATA AS A SPECIAL CASE OF MIR. V. Ramakrishnan^1,2, V. Biou², F. Shu², and R.M. Sweet², Department of Biochemistry, University of Utah School of Medicine, Salt Lake City, UT 84132 and ²Biology Department, Brookhaven National Laboratory, Upton, NY 11973

We show that MAD data can be treated as a special case of MIR, and phased using a variety of popular MIR phasing programs. In this approach, data for one of the wavelengths is considered the "native" data, and the others are considered "derivatives". The dispersive differences between wavelengths are equivalent to isomorphous differences. Bijvoet differences at each wavelength are also included in the phasing, resulting in "derivatives" that have both isomorphous and anomalous differences. In the case of the C-terminal domain of initiation factor 3, three wavelength MAD data were collected on crystals of the selenomethionyl protein. All the MIR phasing programs tried gave easily interpretable maps that had excellent map correlation coefficients with the final structure. We also discuss our experience with a number of issues such as which wavelength should be considered the reference or "native" data set, how additional MIR phase information from heavy-atom derivatives should be included with MAD data, etc. Results on data where MAD alone failed to produce interpretable maps will also be discussed.