I am trying to run an atomistic MD simulation (CHARMM36 FF) for a lipid-peptide assembly, where the peptide is kept on the top of an atomistic bilayer. The peptide is linked to one of the lipid via a linker (I parametrized the linker and the end residue of the peptide). After 30 ns of MD run, the simulation crashes and it generates the PDB structures of the final frame where the covalent bonds in the molecules are broken. This breaking thing is not necessarily limited to the part which I parametrized myself, but mostly some other parts of the peptide breake as well. I have tried to run the lipid-linker-peptide molecule in vaccum just to see if there is anything wrong with the parametrization/topology, but this runs fine for 300-400 ns. But as soon as I solvate it (with water) and then run the simulation, the peptide part breakes and simulation crashes. Any idea?