Using tauRAMD for binding

[FarzinSohraby]

Hi

I was wondering if one can use groamcs-ramd to get binding events instead of unbinding. Is it possible to change the parameters in which the ligand approaches the binding site instead of getting away from it?
The "ramd-group1-max-dist" parameter is for the maximum distance and after this the simulation stops. Can we reverse this to see binding events?

Thank you in advance

Best,

Farzin

[AegithalosCaudatus]

Hi Farzin,

Reversing the ramd-group1-max-dist will not work because COM-COM distance is compared to this value and simulation is terminated when the COM-COM distance > ramd-group1-max-dist.

Regarding your question about the binding events, it is possible, but it depends what are you aiming to do. RAMD will make the ligand move from the positions it has visited. If you are starting away from the binding site, you can approach it by chance when using RAMD (and it should approach and sample the states way quicker compared to using standard MD). It is possible to use some biasing approach in conjunction with RAMD to approach the binding site, since the RAMD will only reply on chance.

It is still needed to keep ramd-group1-max-dist high when using RAMD in this approach, because you still want your simulation to terminate in case your ligand moves too far from the protein.

Hope it helps,

Best,
Mislav

[FarzinSohraby]

Thank you Mislav

I am trying to develop a method to capture binding events and I was thinking if it's possible to reverse the pull code use in TauRAMD for binding. It is great that you think it's possible but I still don't know how to adjust the TauRAMD parameters to make the ligand approach the binding site.

Do you have any idea by chance?

Thank you in advance

Farzin

[AegithalosCaudatus]

Essentially, RAMD enhances the moving of the ligand from the current position and, as you know, enhances the sampling of unbinding events.

If you are starting from a structure where ligand is outside the binding pocket and you are interested in capturing this binding events, you are replying on a random force to approach you to the binding site.

Question is do you already know the location of this binding site? If yes, you would need to use some kind of enhanced sampling approach (for example, targeted molecular dynamics) to approach this binding site.

Currently, RAMD parameters that we have cannot be adjusted so that you specifically approach the binding site because of the random unbiased force (for that you would need enhanced sampling approach that I mentioned).

Best,
Mislav