We ran the Phenix map_to_model tool in two settings.
The default setting is to provide the same EM maps as input as CryoREAD (orange
data points, Phenix). We also provided segmented regions of the map, which are
voxels that are predicted to include nucleotides, i.e. sugar, phosphate, or
bases (blue data points, Phenix (Mask)). The benchmark dataset with 68 maps were
used for these plots. a, backbone recall. b, sequence
recall. Other comparison data are provided in Extended Data 10 and Supplementary Table 6.