search

choderalab / yank

An open, extensible Python framework for GPU-accelerated alchemical free energy calculations.
http://getyank.org
MIT License
177 stars 70 forks source link

yank analyze error #1203

Open jslim-furame opened 4 years ago

jslim-furame commented 4 years ago

I did 10 ns GBSA simulation (number of iterations = 10000, 1 iteration = 1ps). The simulation completed successfully. However, yank analyze gives the following error

pymbar.utils.ParameterError: Warning: Should have \sum_n W_nk = 1.

I read some discussions posted in pymbar GitHub, but they are very ambiguous. So, I tried my own solution and commented out some lines in /pymbar/utils.py.

(Following the message on the screen, I commented out "line 360, in check_w_normalized" in utils.py)

Then, I obtained

WARNING: Did not converge to within specified tolerance. max_delta = 2.240191e-02, tol = 1.000000e-12, maximum_iterations = 250, iterations completed = 249 A squared uncertainty is negative. Largest Magnitude = 0.044772 A squared uncertainty is negative. Largest Magnitude = 0.038710 Free energy of binding : 5.538 +- nan kT (3.302 +- nan kcal/mol) DeltaG complex : 68.781 +- nan kT DeltaG standard state correction: -14.264 kT DeltaG solvent : 60.055 +- 0.006 kT

Enthalpy of binding : -56.195 +- 4.131 kT (-33.501 +- 2.463 kcal/mol)

From this, I think I can trust "Enthalpy of binding", but "Free energy of binding" is questionable. What can I do? What is your suggestion?

Thanks in advance!

andrrizzi commented 4 years ago

Hi @jslim-furame . I'd try downgrading to pymbar 3.0.3 if you haven't done so already. This seems to have solved the issue for one of the users (see choderalab/pymbar#320).

jchodera commented 4 years ago

@andrrizzi : Should we pin YANK's recipe to pymbar==3.0.3 until @mrshirts figures out who broke pymbar and why?

andrrizzi commented 4 years ago

I think that is a good idea.