Evaluation of difference maps

The attached script uses Spider and the diffmap program to calculate a difference map between two input maps and the estimated standard deviation of the difference map. The standard deviation is important for determining the significance of the difference peaks in the difference map. For example, to display difference peaks that are above a 3 sigma significance level, one would chose a threshold (for example, in Chimera) that is 3 times the value of the standard deviation of the difference map calculated by the script. To calculate the standard deviation, the input maps must be given as the half-data maps typically used to calculate FSC curves. Therefore, four maps have to be supplied on input (see top of the script file). The script assumes that these maps are provided in Spider format and that all volumes are perfectly aligned with each other. On output, the script provides the standard deviations of the first input map (after summing the two half-data volumes), the second map – scaled against the first map, and the difference map. It also writes out the volume files for all three maps. These can then be displayed with appropriate thresholds, for example using Chimera.

A set of test files and result files are also attached in a tar archive.

Caution: The estimation of standard deviations relies on unbiased half-data volumes. If there is strong over-fitting of noise, the estimated standard deviations will be too small, as will be the threshold suggested for the difference map. This will be apparent when displaying the difference map with a 3 sigma threshold as there will be too many peaks and the volumes of the peaks will be far too big.