First of all, try to be specific about the purpose of the simulation, as you mentioned some of the goals of MDs but there so many other aims for simulating biomolecules, as you already mentioned docking so what are the biomolecules (Protein-protein) (Ligand-Protein) and so on, as this will help you to determine the force fields and how are you going to analyse these data, for instance if your protein is a glycoprotein you may need to used specific force fields to account for these moieties, after that, you can pick up you favorite software (Gromacs, NAMD or desmond) then fetch for force fields that are suitable for your entities, for instance for solvating your protein target and your ligand, or for solvating chains of antibody and that of an antigen and so on.

At the beginning, following a tutorial is very good to gain more experience and get a taste of the normal pipeline of MD

http://www.mdtutorials.com/

Know that MD simulations require a lot of computational power, that'w why they must be done in a workstation that is GPU-powered as well. because you can run anything on your local machine but for instance simulating a 10 ns on a non-GPU powered average CPU machine could be extensively time consuming, so better if you have a separate workstation, better if you have a linux environment as most of these programs are fully operational on linux like gromacs and desmond, but you can always find good alternatives or compiled versions for windows.

Garbage in, garbage out.

Hi Mohammad Tarek ,

First, thank you very much for making this thing clear and thanks also for taking all this time answering me, I welly appreciate this, second, I'm working on Protein-ligand docking using GOLD and Autodock vina soft. I'm trying to do MD simulation for my final screening step for my ligands with high affinity to my receptor I read that this step is very important to confirm my results do you agree with me, sir?

Dr Abderahmane Linani , Yes MD simulations adds more validity to the docking, but still it is not as accurate as binding assays, however, If the procedures are done in an optimum way, a lot of articles have shown the superb supoerposition between both approaches.

I would simply suggest to use Gromacs, following this tutorial

http://www.mdtutorials.com/gmx/complex/index.html

The pipeline is pretty clear and it is more or less generalizable for most of simulations, If you have good computational power, I would recommend a 20 ns simulation, if not, 10 ns or at least 5 ns could give you a minimum view of the complex stability.

Using molecular dynamics simulations, the protein-ligand complex stability can only be determined by comparing solvated protein-ligand complex with solvated free ligand. After all, the protein-ligand binding free energy is the quantification of the relative probabilities of the ligand in the two environments . . .

Hi Mohammad Tarek,

Thanks again for this information Dr. Mohamed, I will try Gromacs using the given tutorial hoping to obtain good results.

Best regards.

Hi Martin Klvana,

Thanks for the response sir, If I want to compare the "solvated protein-ligand complex with solvated free ligand " as you mentioned above which tools/tutorials do you recommend.

Abderahmane Linani, the two systems must be treated in the same way to obtain comparable results. The simplest, but also grounded in Physics and very intuitive, way to estimate the protein-ligand binding free energy is using the linear interaction energy (LRA) and linear response approximation (LRA) methods as implemented in the Q molecular dynamics simulation package (Aqvist): http://xray.bmc.uu.se/~aqwww/q

Note: With LIE and LRA there are only two contributions to the protein-ligand binding free energy (Coulombic and van der Waals interaction energies); entropy is included in these contributions implicitly (ensemble)..

MD simulations are helpful to elucidate protein-ligand interactions. However, you should be careful when using docking poses as input, as these can lead to several artifacts. Simulations need to sample proper phase space in order to yield useful results. Thus, using conventional methods 10-20 ns runs are rarely enough to achieve this.

For further development you should check the following:

10.1016/j.str.2009.02.010

10.1021/acs.jcim.7b00412

10.1039/c9cp02808k

Just as Mohammad Tarek mentioned you should stick to Linux to save yourself from extensive troubleshooting. For hardware, you should use a proper workstation, depending on the software of choice you may use GPU exclusively. Still, a pairing of powerful CPU and GPU is advisable for the use of open-source programs (GROMACS or OpenMM for example).

Hope this helps, regards

Hi Martin Klvana,

Thank you for guide me in this regard, I will try the Q package the manual is very clear.