Hi Yang,
I'm glad you sorted out the D-amino acid situation (for the record, the standard PDB 3-letter codes for D amino acids are "D{first two letters of standard residue code}"). I haven't had much reason to work with these in the past, but trying it now with an example (9f01... it's amazing how much the cartoon for a D-protein sets off my "uncanny valley" vibes) once you have them named correctly things mostly work, but right now D-disulfides do not. Will have to think about the best strategy for that.
Regarding the Lys-Asp isopeptide... adding support for new user-defined residue-residue bonds is something I haven't had the chance to do (partly because of the vagaries of life/career, and partly because I've been watching the development of new force field approaches that would - once mature - remove the need for explicit residue templates entirely). Eric had this well worked out as an automated process in Chimera (in essence the workflow involves creating a model small molecule covering the two linked residues plus some caps to replace the residues up and down the chain, parameterising that using ANTECHAMBER, then applying the parameters back to the original residues in your model). Question for Eric: does the Chimera version write out any AMBER parameter files (i.e. .mol2/.frcmod) that could be used to make the corresponding OpenMM parameter files?
Best,
Tristan