Hello!
What is the recommended way to choose collective variables / reaction coordinates to study ligand binding to a protein (e.g., for PMF free energy profiling)?
I often see the advice like "Do not just use the distance to binding site", but why is it bad and what should be used instead?
Thank you!
The distance to the binding site is useful if the binding/unbinding pathway is simple. It is often used and should not be called as a bad CV.
Sometimes, if the binding/unbinding event is a bit complex, another CV could be added to differentiate different conformational states in the binding/unbinding pathway, to make the convergence faster and make the description of binding/unbinding events more appropriate.
Sometimes, the distance between the ligand and some other parts of the protein is used as CV, instead of the distance between the ligand and the binding pocket. This is due to the fact that the binding site could experience significant conformational change and the distance may be inappropriate to describe some intermediate states.
Personally, simulating protein-ligand binding could be performed with a proper selection of CVs. You need many tests to see which CVs perform well and refine your CV set repeatedly. Some dimensionality reduction method could also help in this process. The starting tests could always use some distances, e.g. the distance between the ligand and the binding site, the number of hydrogen bonds, or the contact number. Then you do some tests and find more CVs. Practical examples could be found in a number of publications and tutorials.
The distance to the binding site is useful if the binding/unbinding pathway is simple. It is often used and should not be called as a bad CV.
Sometimes, if the binding/unbinding event is a bit complex, another CV could be added to differentiate different conformational states in the binding/unbinding pathway, to make the convergence faster and make the description of binding/unbinding events more appropriate.
Sometimes, the distance between the ligand and some other parts of the protein is used as CV, instead of the distance between the ligand and the binding pocket. This is due to the fact that the binding site could experience significant conformational change and the distance may be inappropriate to describe some intermediate states.
Personally, simulating protein-ligand binding could be performed with a proper selection of CVs. You need many tests to see which CVs perform well and refine your CV set repeatedly. Some dimensionality reduction method could also help in this process. The starting tests could always use some distances, e.g. the distance between the ligand and the binding site, the number of hydrogen bonds, or the contact number. Then you do some tests and find more CVs. Practical examples could be found in a number of publications and tutorials.
If the migration of the ligand from the bulk solvent to the buried binding site through the protein matrix or via a tunnel is not the rate-limiting step in the binding process, it is irrelevant to the binding free energy, which can be then calculated from the effect of solvent substitution, i.e., by simulating the protein-ligand complex in solution and the free ligand in solution (linear interaction energy, linear response approximation, free energy perturbation).
Hi. The Pathways of ligand unbinding can be useful To detect Potential Pathways of binding and the correct cv. As a First guess you might use a Partial tempering of the heavy Atoms of the ligand in its bound state with a temperature below the denaturation temperature of the Host .lt. 350 Kevin. I would Run approx. 10 Simulations with different Seeds for the initial velocities. Then you might detect Potential Pathways of binding. I would Run each Simulation longer than 300 ns. The resulting cv can differ largely from a simple translational degree of freedom. Very likely There will be rotational CVS but potentially also changes within the Host Protein.
- Badges
- Science topic
- Similar topics
- Chemistry
- Theoretical Chemistry
- Computational Chemistry
- Molecular Dynamics