E0046

Three Beam Interference is a Sensitive Measure of the Efficacy of Macromolecular Refinement Techniques. Robert M. Sweet, Alexei S. Soares, Donald L.D. Caspar, Edgar Weckert, Annie Héroux, Kerstin Hölzer, Klaus Shroer, Dieter Schneider, & William Nolan, and Brookhaven National Lab, Upton, NY 11973.

A tool is presented for measuring the usefulness of refinement protocols. Triplet phases measured by the three-beam method from insulin crystals were used to measure the improvement of phases during model refinement, and to quantify the contribution made by each step in the refinement. Conventional amplitude data to 1.5Å resolution were recorded from rhombohedral pig insulin crystals using 1.54Å CuK_α radiation and the refined phases were compared to 804 triplet phases measured using a three-beam interference technique from similar crystals using 1.1Å wavelength synchrotron radiation. The average measurement error in the triplet phases was estimated to be 10°.

An initial atomic model and starting phases were obtained from a published structure, and the atomic model was refined against the amplitude data using CNS. The solvent region was improved further using density-modification. Calculated triplet phases were obtained from the model after each step in the refinement and were compared to the recorded triplet phases. The average difference between the recorded triplet phases and the calculated triplet phases was used as an unbiased measure of the correctness of the model at each stage in the refinement. The average discrepancy between measured and calculated triplet phases was reduced from 53.8° to 23.0° in our refinement.

Each triplet phase has a contribution from three different individual phases. The average individual phase error was estimated after each refinement step by subtracting the 10° experimental error, any by assuming that each individual phase contributed equally to the triplet phase discrepancy. Conventional atomic refinement of an approximate starting model reduced the average individual phase error from 30.5° to 15.3°. Improvement of the solvent region, including the density modification procedure, reduced the individual phase error further by 2.7°. Modeling the discreet disorder of four amino acids accounted for an additional 0.6° improvement, and the final individual phase error was 12.0°.

Our objective is to use these remarkably accurate triplet phase data to determine the relative effectiveness of different refinement protocols, and we invite collaborators to use our data to validate refinement methods which they are familiar with.