An automated peak fitting procedure for processing protein diffraction data from a linear position-sensitive detector

A data processing method is presented which allows reliable intensities to be obtained from the weak neutron diffraction produced by protein crystals. Three components are used to reduce the errors in the intensities: (1) a pattern-recognition technique to find the peak centers; (2) a position-sensitive linear detector, whose spatial resolution characteristics increase the accuracy with which the boundary between the peak and background may be defined; and (3) an averaging technique to obtain accurate backgrounds. Test results with a 2.2 Å data set collected from a small crystal of the proteolytic enzyme trypsin indicate that data processed by this method are superior in both reproducibility and statistical reliability to the same data processed by more conventional methods. Refinement results with the trypsin structure indicate that the data are accurate as well.