search

Gallicchio-Lab / AToM-OpenMM

OpenMM-based framework for absolute and relative binding free energy calculations with the Alchemical Transfer Method
Other
118 stars 31 forks source link

Disconnected alchemical path #47

Closed egallicc closed 11 months ago

egallicc commented 1 year ago 
          I have a few more questions for your guidance.

fig.zip

These two figures are ligands from a certain experiment I conducted, and their ABFE matches the experiment very well with an error of only 0.02. However, what I want to know is that at 150kcal/mol, the second figure does not perform very well, so is this an occasional match?

The second question is, does the 150kcal/mol-200kcal/mol in the second graph not overlap, indicating that the second path is not very suitable? If there is a need to adjust the path, is there a better adjustment direction?

Thank you very much.

Originally posted by @9527567 in https://github.com/Gallicchio-Lab/AToM-OpenMM/issues/42#issuecomment-1788311206

egallicc commented 1 year ago

The sequence of probability densities of the perturbation energy indicates that the alchemical path is disconnected. There are regions where the probability densities do not overlap. The resulting free energy estimate cannot be trusted.

We often see this behavior when the perturbation is too large. In our experience, ABFEs with ligands with more than 10-15 heavy atoms fail in this way. Unless there are other issues, I recommend breaking up the binding into smaller steps. Here is an example.

9527567 commented 1 year ago

Thank you, I will try.