yank
yank copied to clipboard
choderalab
difference in experimental and calculated binding free energy for p-xylene -T4Lysozyme example
I tried the example of absolute binding energy calculation of p-Xylene to T4-Lysozyme in explicit solvent with the yaml file provided in the website withot any changes. My results show free energy of binding as -10.22 +- 0.7 kcal/mol which is far different from the experimental binding energy of -4.67 kcal/mol. Unable to trace out the reason for this difference. I presume that the calculations are hardware independent. Can someone help me to figure our the error
If you could upload the analysis notebook generated using this command: http://getyank.org/latest/analysis.html#visual-analysis we might be able to spot what is giving you trouble.
@mikemhenry Thank you for your immediate response. Please find the analysis notebook here p-xylene-lysozyme.pdf
I would try running longer, it looks like the complex phase took awhile to equilibrate,
so most of the sampling data is discarded. Try running 1000 iterations instead of 500 and see if that improves things.
I have continued my simulation upto 700 iterations initially and thought to further continue this to 1000 iterations as suggested. To my surprise, the free energy of binding after 700 iterations is shown as -17.55 +- 0.7 kcal/mol which is again far different from the experimental binding energy of -4.67 kcal/mol. The system got equilibrated somewhere between 100 - 200 iterations.
Im surprised to see this trend and curious to find out if it is due to some error. Also, i hardly understant how the equilibrated blue line shifted from around 400 iterations (in the initial run with 500 iterations) to 200 iterations (intial 500 extended to 700 iterations)
Please check the notebook generated
p-xylene-lysozyme_700.pdf
Thanks for giving this a try and reporting back! Did you use this example?
Our previous work has required something like 40 ns/replica (40000 iterations) to produce reliable binding free energy estimates in explicit solvent. There is a trend we note in this paper is that the binding free energies start off very negative and converge closer to the asymptotic limit from below with replica exchange.
Absolute free energy calculations are still a very active topic of research, and YANK should be considered a highly experimental framework for protein-ligand binding free energies at the moment.
Thank you for the clarification. I used the p-xylene binding in explicit water.