I was just wondering how far one may go with the pKa predictor and got myself a mini-protein from the pdb (see attached .sdf obtained via molconverter from the downloaded pdb). Then I tried my small hydrogen-setting prog setH.java, which works fine for small molecules.... and KRASH (java.lang.RuntimeException: Molecule@787171 is not a parent of child S-group SuperatomSgroup@15fea60 - I'm still in v 3.5!)! It breaks at plugin.setMol(inMol) - is it too big? Can this be fixed easily? I don't expect the pKa plugin to account for through-space interactions and dielectric effects, but to recognize lysines, arginines and COO- and set their protons properly (that's more than many other programs are able to!) My goal is to later teach the routine to recognize "false" C and N-termini resulting from cuts through the peptide chain around the active site and NOT to charge those atoms, while nevertheless charging the sidechains and the real N/C termini. Feasible?