Re: [AMBER] Help needed with thermodynamic integration: Collision of cations in binding energy calculation

From: He, Amy via AMBER <amber.ambermd.org>
Date: Sun, 19 Mar 2023 14:25:01 +0000

Hi Kevin,

I found on page 450 of the Amber20 manual
https://ambermd.org/doc12/Amber20.pdf

that Table 23.3 provides a summary of how it works for nonbonded interactions within the softscore potential. If my understanding is correct, in previous versions of amber as well as by the default option, the interactions between the softcore atoms are not scaled. This might explain why the softcore atoms are pulled closer than expected especially in the case of ions. But there seem to be other gti_add_sc options to alter scaling schemes if you run TI with the pmemd.cuda.

I use sander.mpi so I think the scaling scheme is consistent with gti_add_sc = 1 for pmemd.cuda..

Bests,


--
Amy He
Chemistry Graduate Teaching Assistant
Hadad Research Group
Ohio State University
he.1768.osu.edu<mailto:he.1768.osu.edu>
From: Kevin Mei <kevinmei2019.outlook.com>
Date: Thursday, March 16, 2023 at 8:29 PM
To: He, Amy <he.1768.buckeyemail.osu.edu>, AMBER Mailing List <amber.ambermd.org>
Subject: Re: Help needed with thermodynamic integration: Collision of cations in binding energy calculation
Hi He, Thank you for sharing your experience. I believe the crux of the problem lies in how AMBER handles non-bonded interactions between two different softcore (SC) regions. This topic is somewhat ambiguous in the AMBER manual. I understand
Hi He,
Thank you for sharing your experience. I believe the crux of the problem lies in how AMBER handles non-bonded interactions between two different softcore (SC) regions. This topic is somewhat ambiguous in the AMBER manual.
I understand that the collision issue only arises when lambda is close to 0 or 1, which means that one SC fragment has minimal interaction with non-SC atoms, allowing it to diffuse freely throughout the system. However, for some reason, the electric repulsive force between two SC regions is even smaller than the van der Waals force, causing the two fragments to collide.
I am unsure if this is a bug or not, but in my opinion, it should be clarified in the AMBER manual how the interactions between two different SC regions are scaled with lambda.
Best,
Kevin
________________________________
From: He, Amy <he.1768.buckeyemail.osu.edu>
Sent: Friday, March 17, 2023 0:27
To: Mei Kevin <kevinmei2019.outlook.com>; AMBER Mailing List <amber.ambermd.org>
Subject: Re: Help needed with thermodynamic integration: Collision of cations in binding energy calculation
Hi Mei,
I am doing very similar calculations as you do, and I have the same problem. I hope others on this mailing list can help us.
I feel the biggest problem is that the cation is 100% mobile (not constrained by any type of bonded interaction) and very skinny, so mine is just stuck in the cluster of softcore atoms. In that situation, when lambda approaches 1, the van der Waals energy becomes extremely high.
I think the transformation of the cation could be buffered by a water. I learned from some people that when they let a Na+ disappear, in exchange they will put a water back. But water is much larger than Na+, so I’m also not sure whether the van der Waals interaction will be underestimated in that way.
I have experimented with many different “transformations” since I need to let a Na+ go for the neutrality of the system.. the van der Waals energy is always problematic for me whenever the ion is involved. But maybe I’m not using the best mixing scheme. My lambda windows are linearly spaced and I use the default alpha and beta in Amber 20 for sander.MPI.
I look forward to hearing from you and others opinions on this..
Bests,
Amy
From: Mei Kevin via AMBER <amber.ambermd.org>
Date: Thursday, March 16, 2023 at 12:06 PM
To: AMBER Mailing List <amber.ambermd.org>
Subject: [AMBER] Help needed with thermodynamic integration: Collision of cations in binding energy calculation
Dear AMBER users,
I hope this email finds you well. I am currently using the thermodynamic integration module implemented in AMBER 20 to calculate the absolute binding energy of a cation to a protein pocket. To achieve this, I am attempting to let one cation inside the pocket vanish while simultaneously letting the one in the bulk appear. I am using a harmonic potential to restrain the pocket-located ion and a softcore potential to smooth the van der Waals energy.
However, I have encountered a problem. When I checked the snapshots of the lambda=0.240452 run, I observed that the two cations have strangely collided together. The cation in the bulk moved inside the pocket, and they are now only 1.1 Å apart. I am concerned that when lambda reaches 1, the ion that should have been in the bulk may become stuck and unable to diffuse into the bulk, which is incorrect.
I would greatly appreciate your help in determining if this phenomenon is normal in TI simulations, or if I have made some errors. Here are the TI parameters I am currently using:
icfe=1
clambda=0.240452
timask1=":913.Na+"
timask2=":914.Na+"
ifsc=1
scmask1=":913.Na+"
scmask2=":914.Na+"
gti_chg_keep=0
ntf=1
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!KGKeukY!yZzyDxTwO3k3FbAAEaBtHZiXD5Vzq7Veq5o5m3rluN6S0I6Z9UYVEfGPy8zmX7LGWSmPOwwy6pzk3ysta7-O3HTLpt4$<https://urldefense.com/v3/__https:/jpn01.safelinks.protection.outlook.com/?url=https*3A*2F*2Furldefense.com*2Fv3*2F__http*3A*2Flists.ambermd.org*2Fmailman*2Flistinfo*2Famber__*3B!!KGKeukY!yZzyDxTwO3k3FbAAEaBtHZiXD5Vzq7Veq5o5m3rluN6S0I6Z9UYVEfGPy8zmX7LGWSmPOwwy6pzk3ysta7-O3HTLpt4*24&data=05*7C01*7C*7C593741439d1d463db54808db263b4b1f*7C84df9e7fe9f640afb435aaaaaaaaaaaa*7C1*7C0*7C638145808340699767*7CUnknown*7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0*3D*7C3000*7C*7C*7C&sdata=*2BUyfkfJtYHaskQJtPUgyDHB8AhYyB2gtjYT*2FYl6*2FHts*3D&reserved=0__;JSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUlJSUl!!KGKeukY!zSWWPMgwAj3Amps_492dtGZ7ctOhFe6JdLCWmAZCLQ4QsiulE4vpSr3AviUAFmJc24LnRf028ba7toMUA2yx5HWyxNpaJsgLpAM$>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Mar 19 2023 - 07:30:02 PDT
Custom Search