custom terms and force field

[Gabriele Palma]

Dear Dr. Koes,
I have started using Autodock Vina to investigate the interaction between a small organic or metallo-organic compound and a target protein, in order to find a potential inhibitor for this protein.
Since I am interested in customizing the minimization parameters to better adapt the calculation to our system and, hopefully, extend our studies, I have found that the program smina appears to be really well suited for this purpose.

Unfortunately, I can not clearly understand the meaning of the custom terms (i.e. what kind of interaction they modify and how) and the parameters structure of the internal force field. I have read your tutorial: "Custom Scoring CDPK1 with smina" but I could not understand several details. I guess the reader is supposed to have a more in-depth knowledge of docking compared to what I currently have.

In particular, I kindly ask if you could indicate where I could find some more specific information/documentation in order to use the custom terms and overwrite the atom types.

Thank you very much indeed for your attention and help

Best regards
Gabriele

--
Gabriele Palma, BSc in Chemistry
Currently MSc student at University of Florence, ITALY

[David Koes]

Other than the README there isn't any more documentation. You can see the implementation of all the potentials here: https://sourceforge.net/p/smina/code/ci/master/tree/src/lib/everything.h

Do you have a specific question about what you want to do?

[Gabriele Palma]

The aim is to simulate some interaction that are not usually considered in docking, defining a specific potential between some pairs of overwritten atom types. I'm trying to understand how I can use smina custom terms and force field parameters to implement this. I would ask what "good distance", "bad distance" and "cap" parameters, wich are found in some custom terms, means and what atom types "depth", "solvation", "volume" and "xs_radius" refers to. Thank you very much for your help.

[David Koes]

First decide what functional form you want for your interaction. The choices for custom atom types are linear, quadratic, gaussian, inverse_power, and lennard_jones. All of these potentials are functions of the distance between the atoms, including their radius. That is, if the distance between the atom centers is 3.0 and each atom has a radius 1.5, the distance (d) used by the potential will be 0. See Figure 2 of this paper (d_diff is what I'm calling d here): https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3726561/

A linear potential is what Vina uses for its hbonds. You specify a good and bad distance and it will linear scale from 1 to 0 between these distances (assuming good < bad).

A quadratic function is the square of the distance. An offset can be provided to shift the optimal location from a distance of zero (when the atoms are touching).

A gaussian is a gaussing centered around the distance (plus any provided offset) with the provided width. That is , exp(-((d+offset)/width))^2)

inverse_power is 1/d^power.

lennard_jones is a 6-12 lennard jones potential. The position of the minimum can be shifted by specify an offset (o), again relative to the optimal distance of zero.

All (most?) potentials take a cap and a cutoff. The cap is a maximum value returned for the potential (to keep things from exploding) and the cutoff is the maximum distance that will be evaluated (any atoms farther apart than this distance will be evaluated to zero).

For atom typing, the xs_radius is the radius used for the above distance calculations. The covalent_radius field is not used for this purpose but for bond determination. This is because we only consider heavy atoms so the effective radius for scoring is set larger than the covalent radius. The other terms (volume, depth, solvation) are only used by the ad4_solvation potential and can be ignored.

If you are defining custom atom types that you want to participate in standard potentials you will want to define the hydrophobe/donor/acceptor information appropriately.

[Gabriele Palma]

Thank you very much indeed for your explanation!