journal menu
![[Figure 5]](jo5030fig5mag.jpg)
|
|
Figure 5
Dependence of SAXS measurement errors on concentration, exposure time and flux. Mean and variances of the buffer-subtracted scattering intensities were determined from repeat exposures under a range of measurement conditions at synchrotron beamline BM29, ESRF, Grenoble (coloured symbols). The absolute variances (a) and the relative errors (b) are shown. Measurement conditions (protein concentration, number of exposures and length of each exposure) were 8 mg ml−1, 5 × 2 s (cyan); 8 mg ml−1, 10 × 2 s (blue); 8 mg ml−1, 10 × 4 s (green); 24 mg ml−1, 10 × 4 s (red); 24 mg ml−1, 10 × 4 s at half beam intensity (magenta). Fits of our error model with c and k as fitting parameters are shown as black solid lines. The fitting parameters are (using qarb = 0.2 Å−1) I(qarb) = 12.21, k = 6134, c = 0.8750 (cyan data); I(qarb) = 24.20, k = 5873, c = 0.8757 (blue data); I(qarb) = 45.77, k = 5273, c = 0.8737 (green data); I(qarb) = 142.08, k = 4822, c = 0.6894 (red data); I(qarb) = 75.49, k = 5507, c = 0.7186 (magenta data). When fitting a straight line to the number of pixels per q bin N(q), a value for k of the same order of magnitude is obtained (k = 4387; Fig. S5). The inset in panel (a) shows the relative error at qarb = 0.2 Å−1 as a function of the (flux-corrected) exposure time. The solid lines are fits of a relationship αtexp−1/2, which provide an excellent description of the data for both the low- (8 mg ml−1; open symbols) and high-concentration (24 mg ml−1; solid symbols) data. From the fit we find α(8 mg ml−1)/α(24 mg ml−1) = 2.7, close to the concentration ratio. The inset in panel (b) shows the ratio of relative errors between 8 and 24 mg ml−1 data at the same texp (green divided by red data from the main panel), which is generally close to 3 (indicated as the dashed line), but varies with q. |
![[Figure 5]](jo5030fig5.jpg)
 | JOURNAL OF APPLIED CRYSTALLOGRAPHY |
ISSN: 1600-5767
access
