Do rotamer libraries reproduce the side-chain conformations of peptidic ligands from the PDB?

Rotamer libraries are collections of side-chain conformations compiled for each type of amino acid. All the libraries developed so far were constructed based on different sets of proteins, none of which includes small peptidic molecules. Are the existing rotamer libraries suitable for modeling the side chains of peptidic ligands in complex with their receptors? To answer this question, we have tested 10 different, publicly available and commonly used rotamer libraries for their capability of reproducing the side-chain conformations (and therefore the receptor-ligand interactions) of hundreds of short peptidic ligands found in the Protein Data Bank, including large sets of class I and class II T-cell epitopes. Only the libraries developed by Xiang and Honig were able to correctly reproduce the experimental geometries for most of the analyzed residues, and the atomic interactions between the peptidic ligands and their receptors. Surprisingly, all the libraries showed a lower performance in reproducing the side chains conformations from structures solved at very high resolution (R<1.25A).