Determination of the Iridium positions in the unit cell by deconvolution of anomalous difference Patterson maps proved to be non-trivial. This was probably due to the presence of multiple Iridium sites, as had indeed been suggested by the initial screening of the derivative. Direct methods were therefore considered as a more appropriate option for solving the heavy atom structure. The program used was MULTAN as described in Sec. . The Bijvoet differences calculated from data set 3 were used along with MULTAN in an attempt to locate the positions of the iridium atoms in the lysozyme derivative.
Bijvoet differences were rejected as input to MULTAN on the basis of two criteria. Terms were rejected if or if . On this basis 704 and 24 reflections respectively were rejected and 2895 used as input to the program. An additional 859 centric reflections were not used since no Bijvoet differences exist for these. A total of 36 phase sets were generated by the program and figures of merit calculated for each.
Table shows the results for those phase sets with a combined FOM greater than 1.5. Phase sets and E-maps were calculated using 200 reflections. Most of the maps contained large spurious peaks at special positions in the cell and were therefore deemed unlikely to be correct. Phase set 15 had the highest combined FOM (2.522) and set 1 had the lowest values of RESID (39.84) and the highest ABSFOM (0.440).
Table: Results of direct methods phasing with MULTAN using Bijvoet differences from data set 3 collected at the Ir white line. A total of 36 phase sets were calculated by the program but only those with combined FOM's greater than 1.5 are shown here. The best phase set as suggested by combined FOM is set 15. The E-map calculated from this set did not contain any peaks which coincided with the known heavy atom positions. All of the E-maps calculated from these phase sets contained one or more peaks which occupied special positions and often were dominated by one large peak which was sometimes four times larger than the next highest feature.
The procedure was then repeated using the dispersive anomalous differences between data sets 4 and 5 at the two inflection points on the white line. The difference between these two energies is and the values at the two points are 12.1 and respectively. In all other respects the data sets are essentially identical due to the very small energy difference between them of .
Using a similar argument to the one above it can be assumed that for the largest values of ,
Using similar criteria as had been applied to the Bijvoet differences reflections were rejected leaving a total of 3655 reflections, including centrics, which were normalised and used as input to MULTAN. A total of 16 phase sets were output and are shown along with there figures of merit in Table . The best combined FOM (2.993) was observed for phase set 5 which also had the lowest RESID (26.74) and the highest ABSFOM (0.623). The E-map calculated using this phase set displayed five large peaks none of which were at special positions. The MULTAN sites are listed in Table along with the respective refined positions
Table: Results of direct methods phasing with MULTAN using the dispersive anomalous differences between data sets 4 and 5 collected at energies corresponding to the rising and falling inflection points of the Ir white line.
The next best solution (set 12) reproduced the second site in Table and gave a peak Å away from the first site. In addition a peak with and coordinates equal to the third site in Table was observed but shifted by along the direction. Indeed several of the phase sets inspected produced E-maps containing one or two correct sites.
Table: The fractional coordinates of the five peaks in the E-map calculated from the best MULTAN solution as judged by the figures of merit. Shown along side are the corresponding refined atomic positions and occupancies of the iridium from the SIRAS study. The four highest peaks in the E-map corresponded to the four most occupied iridium sites.