CHARMM Development Project Forums User Discussion & Questions Parameter Set Discussion Use of different versions of parameters

I have noticed that certain vdw parameters such as those of alkanes changes (or rather improves) from one version to another and different atom types were defined. For instance, CT3 for protein and CTL3 for lipid.

The CTL3 is being used in lipid section, while the CT3 is used in the protein section of par_all27_prot_lipid.prm. This is also the case for those fitting molecules in toppar_all27_lipid_model.str and top_all22_model.inp respectively.

My question is that is it okay to substitute CT3, HA etc in protein parameters for CTL3, HAL3 etc in lipid parameters? In other words, change all CTs and HAs in amino acid residues for CTLs and HALs. For instance, methylacetate were both defined in both toppar_all27_lipid_model.str and top_all22_model.inp. Both are identical in terms of charges and the only difference is the use of CTL3 and CT3 which have slightly different vdw values. Which one should I use?

For amino acids, peptides, etc., it's probably best to stick with the protein parameter set chemical types, as the angle and dihedral terms are tuned to the VDW values.

Then how about, say, modelling of alkane chains with proteins?

For instance, shall I use, say, butane and pentane as defined in toppar_all27_lipid_model.str or those alkanes as defined in toppar_all22_prot_model.str, in conjunction with proteins as defined in top_all22_prot.inp ?

I would tend to use the protein force field for anything that's covalently bound to proteins (and reasonably supported by the protein FF). This will spare you a lot of "missing parameter" headaches and inconsistencies. For free alkanes and esters, however, it's probably better to use the lipid force field. The model compounds in the protein force field were parameterized with the goal of representing proteins, while the model compounds in the lipid force field were parameterized mostly with the goal of representing "free" alkanes/esters...
The nonbond interactions are supposed to be consistent between CHARMM force fields in the sense that you can use different CHARMM force fields (protein, lipid) for different molecules in the same system, as long as they're not covalently bound (typical example: a transmembrane protein in a lipid bilayer). If they are, it usually is still possible to combine them, but it becomes more tricky and may require re-optimizing dihedrals...