Figure 2.

Merging statistics for a hen egg-white lysozyme sulfur SAD data set. (a) The Python function for applying inverse-variance weights to obtain maximum-likelihood merged intensity estimates using reciprocalspaceship. (b) Correlation coefficients, CC_1/2 and CC_anom, from repeated twofold cross validation. The Pearson CC_anom is more affected by outlier measurements in low- and intermediate-resolution bins than the Spearman CC_anom. (c) The distribution of residuals from observed intensities differs from the expected distribution of residuals for normally distributed observations. (d) CC_anom from twofold cross validation using Student t-distributed error models with varying degrees of freedom (d.f.). Error models with heavier tails show improvements in CC_anom. Error bars depict the mean ± standard deviation from 15 repeats of twofold cross validation.