ATB server

GROMACS version: 5.1.5
GROMACS modification: Yes/No
Dear Justin,
I have used the topologies (GROMOS54A7) generated directly by the ATB server to simulating a cyclodextrin/drug complex, without any additional modifications. Is this process correct and true? Is it correct to say that I used GROMOS54A7?
Many thanks,
Ganj

I haven’t used GROMOS force fields for anything in years, but ATB is (as far as I know) the most up-to-date and reliable method of generating GROMOS-compatible topologies for arbitrary species.

Dear Justin,
Many thanks for the answers.
Best regards,
Ganj

The ATB should be fine for your drug although, as with any of these automated servers, it may be prudent to do some further validation on the parameters that are generated.

As for GROMOS and cyclodextrin, you’d be better off using some refined/validated parameters of the carbohydrate force field. In particular, have a look at the following paper for a comparison of a few force fields (including several GROMOS ones):

https://pubs.acs.org/doi/10.1021/acs.jpcb.7b11808

I’ve got a topology for (IIRC) beta cyclodextrin with the GROMOS 53A6GLYC parameters somewhere that I can dig out and send, if that helps?

Cheers

Tom

Dear TomPiggot,
Many thanks, I also read this article “https://pubs.acs.org/doi/10.1021/acs.jpcb.7b11808”, but now I want to be sure about just ATB server. I want to be sure that this setup is true and reliable. Would you please send me that topology file? How can I build a topology for drug based on this force field? Is it true that I used ATB server for both of them without any corrections?
Best regards,
Ganj

Hi,

Before I answer, I’m going to make a couple of suggestions (obviously feel free to take or leave them as you see fit). Firstly, I’d suggest that you have a good general read about different force fields before you get started and be sure that you want to use a GROMOS force field for your simulations. I’m not saying don’t use a GROMOS-based force field, rather I would just recommend that you (and indeed anyone doing simulations) know why you have chosen a particular force field to simulate with and why (that’s if you decide to only do simulations with one force field). Secondly, if you do decide to go with the GROMOS force field family, I’d strongly recommend that you have a read of some of the GROMOS papers before you get started. There are several good ones out there including the GROMOS 53A6, 54A7, 54A8, and 2016H66 publications; several detailed reviews, especially those from Wilfred van Gunsteren and co-workers; there are also a couple of papers from the group of Alan Mark regarding the ATB and how it works, etc.

Anyway, as to your questions, the ATB should be fairly reliable for your drug. However, without knowing the structure of your drug it is hard to say. Plus it is up to you to ensure that the parameters the ATB assigns (which will mostly be based upon analogy to the existing force field, plus using QM for the charges) are ‘accurate enough’ for your required purpose. What ‘accurate enough’ means is up to you to decide but you will need to be able to justify whatever you have done at a later date to a reviewer, if asked. Typically for GROMOS force fields, validation is done using things like solvation free energies and so forth (the papers I mentioned above have far more detail on this). That said, there might well not be such experimental data for your compound out there meaning this may well not be possible. You may also want to do things like performing torsional scans to ensure you’re getting a close enough match to QM data, for example. It’s also worthing point out here that other force field families (e.g. CHARMM) have much more defined processes for both assessing and refining such automatically generated parameters.

With regards to using the ATB output with the rest of the GROMOS 54A7 force field, there shouldn’t be any problems. The ATB sometimes does less standard things with some molecules (e.g. including some additonal explicit hydrogen atoms that you wouldn’t normally get with a united-atom force field) but there are force field files on the ATB website you can download and use for these cases.

With regards to the cyclodextrin, you shouldn’t use the ATB to generate your parameters. The ATB is designed to produce general parameters for small molecules using a generic workflow and so should not be used for molecules which have been extensively paramaterised and validated using other, typically more hands-on, approaches. For the cyclodextrin, and as per the paper I linked to before, you should use one of the GROMOS carbohydrate force fields that have been validated for cyclodextrins (e.g. GROMOS 53A6GLYC). The ATB parameters for the drug and the cyclodextrin parameters should be compatible with one another.

Hopefully that all makes sense. I’ll look out the cyclodextrin topology later and send it over. I’ll note now that I made this a while ago for someone else and so you should check the topology yourself against the GROMOS 53A6GLYC paper; I haven’t performed many simulations with it myself.

Cheers

Tom

Can we use the all atom itp file given be atb server in any other force field which uses all atom model?

No. The parameters produced are not transferable to another force field. And given that the GROMOS force fields are united-atom, this all-atom topology should not typically be used.

That said, there are plenty of other similar websites out there for all-atom force field families (e.g. LigParGen for OPLS-AA; ParamChem for CGenFF/CHARMM; GAFF or GAFF2 and several websites for AMBER) which will (well in the majority of cases) use appropriate force field specific methods to generate parameters for small molecules (e.g. by analogy to the rest of the force field, or through applying the appropriate QM methods, etc.). However, as per my previous message, validation and maybe further refinement of these automatically generated parameters is advisable. ParamChem with CGenFF is especially useful here as it gives you penalty scores for the parameters it assigns by analogy.

Cheers

Tom

1 Like

Yes I get it. I was just curious why give an all atom file when their is no use of it.

I’ve attached an itp and gro for beta-cyclodextrin using the GROMOS 53A6GLYC force field (and which can be used with the GROMOS 54A7 force field files provided with GROMACS). As I mentioned, it’s worth double checking what I’ve done as I’ve not extensively used this.

Cheers

Tom

(Attachment bcdex_53A6GLYC_em.gro is missing)

(Attachment bcdex_53A6GLYC_explicit.itp is missing)

It appears only certain types of attachment allowed to the new list. I’ve renamed the files and am trying again (you’ll need to rename the top to itp and the dat to gro). If that doesn’t work, dr_m_ganj please send me a message off-list and I’ll send you the files.

Cheers

Tom

bcdex_53A6GLYC_em.dat (4.41 KB)

bcdex_53A6GLYC_explicit.top (32.5 KB)

Dear TomPiggot,
Many thanks for your attentions, I used all of your comments and they were useful to me. If I do validation for the beta-cyclodextrin (by the files that obtain directly from ATB), can I used it directly? (For both of them). Many thanks for the files.
Best regards,
Ganj