The data sheet produced affinities of 4800 consecutive segments of 15 amino acids, each overlapping by one single amino acid progressing downstream to cover the entire length of the 188 AA mucin-like domain of the Ebola Glycoprotein.
manually, you could load the structure in a 3D viewer and colour your best assay-coring peptides to see which are co-localized in 3D (I used rasmol in my time do do this).
I am sure there are programs to do this on a more automatic way, based on the closeness (distance) between aminoacids side-chains from the assay-scoring peptides and exposure to the solvent.
Bottleneck is that you score from linear peptide scans and will not catch conformational epitopes.
The modeling of that region with the exception of the E1 epitope is that its in the mucin-like domain and refractory to crystallography. Modeller is not helpful for that reason. I'm trying to demonstrate the regions that negatively correlate with Reston as compared with the virulent species.