AMBER: About MM/PBSA, PB Calculation

From: nlxc <>
Date: Wed, 15 Dec 2004 15:11:29 -0800 (PST)

I am trying to validate my MM/PBSA protocol and AM1-BCC charge by calculating the hydration energy of about 400 small compounds. Generally the correlation is ok, but those high polar ones have large errors. The energy difference between experiment data and my results of some compounds are as large as 10-15 Kcal. My calculated results for most of the compounds shift to the more negative region comparing with experiment data. I use Delphi for PB calculation. Ligand charges from AM1-BCC has been added into Delphi input file.

1) I found almost all of those compounds with large error are compounds with amine group (NH2 in the terminal.) or group O-C=O or group H-C=O Almost all the compounds with the above groups are not calculated very well.

2) I checked the plots of RMSD vs. time and PBCAL vs. time, most of those "bad" compounds are not very stable, but not particularly unstable comparing with the rest of "good" compounds. Lot of them flucatuate a lot ( from 0.2 - 1.0 A RMSD). But there are also several "bad" compounds with very stable RMSD and energy (PBCAL) plots.

Since I have run about 1ns MD for those compounds with large error, I don't think simple longer simulation with the current protocol can make them more stable. I have also checked all the bond orders. Since they all are neutral compounds, the charges added from prep file are almost 0 (~0.004).
I added about 10A TIP3 WAT BOX around each ligand, run NVT first, then NPT for total 1ns.

Please let me know if you have any suggestions about why those particular subgroups affect my PB calculation and how I can improve my results. How to improve my MD protocol to make the those small ligand simulations more stable?

Thanks a lot. Happy Holidays!

Do you Yahoo!?
 Yahoo! Mail - now with 250MB free storage. Learn more.
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Dec 15 2004 - 23:53:01 PST
Custom Search