I'm working on de novo enzyme design. Utilizing the enzdes application through theozyme definition. While doing regular design procedures I have noticed that Rosetta inserts amino acid identities that create clashes with the ligand. Most notable is an Ile with a distance of 3A to the ligand, instead of the minimum 4. Images attached. As far as I know, I have not altered the default behavior of vdw radii at any point. The energies for these residues do not sky-rocket and Rosetta scores these designs very favorably.