Rosetta’s actual complaint is that it does not recognize the residue code field (CH), not HETATM/ATOM. As a general rule, you should convert HETATM to ATOM anyway.
What is a CH supposed to be? I assume it’s a modification of one of the four standard bases? If you remove the CH lines, it won’t fail on them, but you’ll have gaps. If you convert it to the closest residue type, then it may have an atom or two wrong but will be mostly correct. If you really want CH, whatever it is, you can create a parameters file for it.
Is 1KPY from the documentation/demo, or the one from the PDB you want to use? Nucleic acid formatting in the PDB is poorly standardized; Rosetta only takes the input in one format, so you may have to use scripts to tweak things (especially the residue code fields) to get Rosetta to recognize it.