MIR: AN AUTOMATED PROGRAM FOR ISOMORPHOUS REPLACEMENT

A.A. Vagin¹ and M.N. Isupov²

¹ Department of Chemistry, University of York, Heslington, York, UK
² Departments of Chemistry and Biological Sciences, University of Exeter, Exeter, UK

MIR is an automated program for heavy atom/anomalous scatterers position location and subsequent phasing of multiple isomorphous and/or anomalous data. It is based on the translation function approach for heavy atom (HA) location in derivative data with anomalous signal as implemented in the program TRAHALO [1]. TRAHALO features simultaneous use of the isomorpohous and anomalous signal for heavy atom location, scaling by Patterson origin peaks and soft resolution cut-off. Initially the HA position search is conducted by molecular replacement in difference Patterson using the full-symmetry translation function (TF) [2] with a one-atom probe model. A number of the translation function peaks which have the highest phasing power (PP) is used for further heavy atom search. The TF search is conducted for each of chosen peaks fixed as a model with known position. The sites thus found are checked against the list of possible heavy atoms from the original HA search. The pair if HA is considered to be acceptable if both HA are present in the initial TF peaks list and there is a cross-validation, i.e. if the second atom is in the list of TF solutions when first is fixed and vice versa. The pair of HA thus found has the common coordinate origin and hand. Among found accepted pairs of HA the one with the highest PP is chosen as a fixed model and the TF search is for the third HA begins. The search for further HA goes on until the addition of the new atoms does not increase the PP.

To make multiple isomorphous replacement completely automated, a 'super' program MIR has been created of which TRAHALO is a part. In case of several heavy atom derivatives MIR uses the HA sites found by TRAHALO for each derivative, calculates cross-difference Fourier syntheses for derivative 2 with phases from derivative 1 (HA-1 sites) and finds a set of HA sites for derivative 2 (HA-2). Using those found HA-2 sites for phase calculation a set of HA sites HA-1-NEW is found in difference Fourier for derivative 1. If those HA-1-NEW sites coincide (at least partially) with original set of sites for derivative 1 (HA-1), HA-2 set of sites is considered to be correct. Thus additional sites can be found for each derivative, some poor derivatives which couldn't be solved by TRAHALO are solved by MIR, and all the derivatives acquire a common origin and hand.

When HA are found for all derivatives, MIR refines their positions and occupancies, calculates phases and does density modification. MIR takes native and heavy atom data as input and produces phases after density modification. Additionally, MIR can use external phases, e.g. the molecular replacement phases.

Recently anisotropic scaling and correction of input data have been introduced in MIR. Although anisotropic scaling of derivatives data to native data is relatively common, native data could be strongly anisotropic as well, which could adversely affect the results of heavy atom location. The work is in progress to determine whether anisotropic correction of native data as implemented in MIR can improve heavy atom search results.

At the moment MIR works with BLANC [3] and CIF [4] input data formats. The work is in progress to make it compatible with MTZ [5] format. The program is available free as part of BLANC from AAV.

[1] Vagin, A.A., & Teplyakov, A. (1998). Acta Cryst. D 54, 400-402.
[2] Vagin, A. & Teplyakov, A. (1997). J. Appl. Cryst. 30, 1022-1025
[3] Vagin, A., Murshudov, G. & Strokopytov, B. (1998) J. Appl. Cryst. 31, 98-102.
[4] Hall, S. (1991) Acta Cryst. A 47, 655-685.
[5] CCP4 (1994). Acta Cryst. D 50, 760-763.