Dear Mudassir,

You asked a strange and interesting question at the same time. In science we do not have absolute proofs and absolute refutations. Not even math provides the proofs we may seek despite the fact that it is the most formal of all. The standard of proof evolves and changes over time. What we use in science is preponderance of evidence. It means that always under some differing set of conditions the result can be different.

In application to you direct question. Both procedures i.e. docking and MD have significant limitations. The list of limitations is so long that there is no room for listing them here. So the positive answer is more of an accident than a method. But obviously both methods provide useful leads.

Therefore, the shortest answer is that you have to provide much more than a simplistic numerical answers in order to "show" sufficient evidence for the binding. The more experimentation the better. Even a single experiment is not sufficient in many cases. If you add on top of it that many, many experiments are impossible to duplicate, you will see how philosophical underpinning of science becomes important in any real life case.

Cheers

Bog

Hi Mudassir, I would probably do a free energy calculation in incorporation with MD simulations in order to validate docking results.

Hi Mudassir Khan ,

In a nutshell, yes, you should combine a different set of analyses to validate your docking results. Here are my suggestions for you to be more confident your docking is OK, using MD:

1) First, you should decide on how long you intend to simulate your protein. The larger the protein, the longer the MD. I would say that if your protein has 400-700 aa, at least 500 ns would be mandatory. 200-400, maybe 400ns or 300ns might be OK.

2) Since you have a ligand, I think that you should use an all-atom force field.

3) I suggest you use triplicates for your system, this way you will be able to isolate low-probability phenomena.

4) After the simulation, measure the atomic distance from your ligand to the amino acids you know are responsible for the binding. This is a simple solution and a very efficient one. This way, you can see, using a simple analysis, if your molecule is distancing from the amino acids as time progresses.

5) (This is *Optional* in my opinion). If your ligand is still in the pocket after 3 simulations, during almost 1us (combined), and it does not distance itself from the pocket after you measure the atomic distance, I would say you are safe. But, if you want to be sure that your docking is OK, like, really sure, you can measure the energy of interaction or perform a metadynamics. I think it would not be necessary anymore.

Hope it helps!

Cheers,

Bruno

Boguslaw Stec Thank so much Sir, your precious time and elaborate answer is very helpful, how beautifully you explained that everything is not perfect ( Both procedures i.e. docking and MD have significant limitations )

Thanks

Inna Ermilova Thank you so much, i would also do the same for validation, hopefully, your suggestion would be helpful.

Thanks

Bruno César Feltes Thank you so much for your good informative and step wise answer.

Thanks

Dear Mudassir,

You got good advice from Inna and Bruno. Unfortunately they are theory practitioners. Free energy perturbation methods are only as good as the potentials they use. The potentials are bad for almost any practical use. A simple look at the bonding term as a harmonic potential and realization that it prevents ionization is sufficient to give you a pause. Despite that they are used quite effectively. You have to learn how and why.

In terms of the length of simulation and dynamical stability it is enough to notice that every single protein is a semi-stable (conditionally stable system). I am not even mentioning many methodological difficulties of MD such as equilibration, thermal control, dynamics versus thermodynamics conversion, time steps size, appropriateness of classical mechanics to simulate quantum systems, etc. etc. Biology tells us that proteins exists on the scale from milliseconds to years and after this period they are destroyed. So looking through the thermodynamic lens at protein stability is completely wrong approach. Proteins exist not through thermodynamic stability but through dynamic stability. Many times when the system is a bit more stable thermodynamically (longer half-life time) your theoretical results would be OK but when system is less stable they cannot. Most of short life-time comes from proteins recycling in cells (active degradation) but many just simply from spontaneous unfolding and subsequent degradation.

Good Luck.

Bog

Boguslaw Stec Sir i am happy with your answers, thank you so much, protein partial stability is fine, if the case is in MD, we can provide periodic boundary conditions to optimize the system and mimic the native protein, if i am right.

Thanks

Hi,

I would like to thanks Prof. Boguslaw Stec for his valuable insight.

Molecular docking is widely used to obtain binding modes and binding affinities of the molecule to the target protein. Despite considerable efforts, however, prediction of both properties by docking remains challenging mainly due to protein 's structural flexibility and inaccuracy of scoring functions. Free energy is the main quantity for explaining the thermodynamics of biological systems. In this context, we consider the measurement of free energy, enthalpy and entropy from the simulation of final molecular dynamics. For higher absolutely we can perform quantum energy calculation or hybrid quantum QM/MM) simulation.

Absolutely, Molecular docking have good initial screening procedure, but MD simulation give overall perception of any complicated thermodynamic system.

Best

Satya

Hi

I would prefer free energy calculation after MD simulation.

Satyaranjan Biswal Thank you for your guidance

Sushma .P Thank you for your valuable response

Very good ques! It can have multiple answers, however people have already discussed about free energy calculations after MD. I would further like to add that, one may choose to perform QM/MM calculations which incorporate DFT and therefore generate reliable results but the level of one is seeking is entirely matter of the project being worked upon.