view article
Figure 1
Bootstrap resampling (Grünbein et al., 2021BB9) by `random drawing with replacement' is used to estimate the effect of measurement errors in SFX data on final refined coordinates. An image is randomly selected from the pool (left) and placed (as a copy) in `resampled' data set 1. The image is then placed back in the original pool, and the pool is then randomly sampled again to add a second image to resampled data set 1. Multiple resampled data sets (up to 100 for example) are constructed that contain the same number of images as the original pool but in which images can be represented multiple times. Structures are determined from each of these resampled data sets, and then refined. The standard deviation of the ensemble-averaged bootstrapped structures gives the mean error of the coordinates. Although computationally time-consuming, bootstrapping can provide a valuable method of estimating coordinate error.

IUCrJ
Volume 8| Part 4| July 2021| Pages 482-484
ISSN: 2052-2525
https://doi.org/10.1107/S2052252521006552