IEDB Analysis Resource

ElliPro - Tutorial

I. Overview
ElliPro predicts linear and discontinuous antibody epitopes based on a protein antigen's 3D structure. ElliPro accepts as an input protein structure in PDB format. If input is a protein sequence, please go to Methods for modeling and docking of antibody and protein 3D structures where information on methods for modeling and docking of antibody and protein 3D structures is available.

ElliPro associates each predicted epitope with a score, defined as a PI (Protrusion Index) value averaged over epitope residues. In the method, the protein's 3D shape is approximated by a number of ellipsoids, thus that the ellipsoid with PI = 0.9 would include within 90% of the protein residues with 10% of the protein residues being outside of the ellipsoid; while the ellipsoid with PI = 0.8 would include 80% of residues with 20% being outside the ellipsoid. For each residue, a PI value is defined based on the residue's center of mass lying outside the largest possible ellipsoid; for example, all residues that are outside the 90% ellipsoid will have score of 0.9. Residues with larger scores are associated with greater solvent accessibility. Discontinuous epitopes are defined based on PI values and are clustered based on the distance R (in Å between residue's centers of mass). The larger R is associated with the larger discontinues epitopes being predicted. See this paper discussing the method performance and the comparison with other methods.
II. How to use the tool
  1. Specifying a protein structure input
  2. Specifying parameters for epitope prediction
1. Input is a protein structure
Input Page:

Figure 1 . Example structure input using PDB ID
PDB Chain Selection Page:
If your protein structure contains more than one chain, ElliPro will ask you to select which chain(s) to be used for epitope prediction. Select one or more chain(s), then click "Submit" (Figure 2 ). IMPORTANT NOTE: ElliPro predicts epitopes for each PDB chain independently, even if more than one chain was selected. Therefore, be cautious when interpreting the results, as the predicted epitope might be buried in the interface with other chains and be solvent unaccessible. If you want to predict epitopes in a multi-chain protein, for example, considering the whole oligomer of the hemoglobin (PDB ID 4HHB, chains A, B, C, D all together), you need to modify a correspondent PDB file (4HHB.pdb) so that all chains of interest have the same chain ID (specified in the column 22 of the ATOM record in the PDB format file). To avoid conflict of residues having the same numbers in the modified PDB file, residues might need to be renumbered.

Figure 2 . Example structure input using PDB ID
3. Epitope prediction parameters
III. How prediction results are presented
An example layout of prediction results is shown in Figure 3.
The results show: Visualization: Placeholder for unknown chain id:
Figure 3. Example prediction results

Figure 4. Example 3D structure mapping of a predicted discontinuous epitope

Figure 5. Example 2D score chart

Figure 6. Example underscore as a placeholder