Re: Dual Topology in FEP Calculations - How to Calculate Electrostatics and van der Waals Free Energies Separately?

From: Nassar, Omneya (omnassar_at_UTMB.EDU)
Date: Tue Jul 24 2018 - 07:29:13 CDT

Thank you Jerome! I appreciate it!!!

Omneya

On Jul 24, 2018, at 3:00 AM, Jrme Hnin <jerome.henin_at_ibpc.fr<mailto:jerome.henin_at_ibpc.fr>> wrote:

With alchVdwLambdaEnd=0 all atoms (outgoing and incoming groups) will be coupled to the environment at all times through the Lennard-Jones potential (lambda_LJ = 1).

I'm now realizing that the transformation you had in mind doesn't work, because it won't sum to the total mutation free energy. What you can do is a series of three simulations:

1 - discharge group A with only LJ of A present
2 - vanish LJ of A while growing LJ of B
3 - charge group B with only LJ of B present

Then the sum of free energies from steps 1 and 3 will give you an idea of the electrostatic contribution, and step 2 gives the vdW part.

Jerome

On Tue, 24 Jul 2018 at 02:54, Nassar, Omneya <omnassar_at_utmb.edu<mailto:omnassar_at_utmb.edu>> wrote:
You are correct; my question is more complex. I am asking about the outgoing and incoming atoms for the alchemical transformation when I have alchVdwLambdaEnd=0 and alchElecLambdaStart=0. So for my system, I am starting with NADH and transforming it to ADP. At the first window (0/32 to 1/32), does the simulation only use the NADH van der waals shape and charges while ignoring ADPs shape for the Free Energy calculation or is the shape of both NADH and ADP used in the calculation and only the charges of NADH are used for the simulation?
Attached is a diagram of what I am asking. If it is still confusing, please let me know. I really appreciate your patience and thanks again for helping me.

Omneya

From: Jrme Hnin<mailto:jerome.henin_at_ibpc.fr>
Sent: Monday, July 23, 2018 4:15 PM
To: Nassar, Omneya<mailto:omnassar_at_UTMB.EDU>
Cc: Namd Mailing List<mailto:namd-l_at_ks.uiuc.edu>
Subject: Re: namd-l: Dual Topology in FEP Calculations - How to Calculate Electrostatics and van der Waals Free Energies Separately?

All particles are present at all times; their interactions vary. I think there was more to your question, but I didn't get it.

Jerome

On Sat, 21 Jul 2018 at 14:59, Nassar, Omneya <omnassar_at_utmb.edu<mailto:omnassar_at_utmb.edu>> wrote:
Thank you Jerome for your prompt response. If you could further answer this question, I will greatly appreciate it. So at FEP lambda 0.0625 to 0.125, if I have 0,0 for alchVdwLambdaEnd, alchElecLambdaStart, respectively, do I have the geometry of NADH and ADP present or do I have the geometry of the intermediate at 0.0625 to 0.125?

Thanks!
Omneya

On Jul 21, 2018, at 6:02 AM, Jrme Hnin <jerome.henin_at_ibpc.fr<mailto:jerome.henin_at_ibpc.fr>> wrote:
Dear Omneya,

Your simulation parameters seem correct to me, as does your understanding of the alchemical process in this implementation.

Best,
Jerome

On Fri, 20 Jul 2018 at 22:07, Nassar, Omneya <omnassar_at_utmb.edu<mailto:omnassar_at_utmb.edu>> wrote:
Hello NAMD staff,

I am trying to properly implement the dual topology paradigm in Free Energy Perturbation technique for a mutation of NADH to ADP. I want to calculate two things: 1) the van der Waals free energy of NADH changing to ADP and 2) the electrostatics of NADH changing to ADP. Thus, I should have values for dGvdw and dGelec from two separate simulations. However, after reading over the documentation (http://www.ks.uiuc.edu/Research/namd/2.7/ug/node53.htmlsearch-252Fnamd-252F2.7-252Fug-252Fnode53.html-26data-3D02-257C01-257Comnassar-2540UTMB.EDU-257C099dbcc3f2c0496789c708d5eef97f81-257C7bef256d85db4526a72d31aea2546852-257C0-257C1-257C636677677711073805-26sdata-3DahfoJSCH4i7e6sVP6dY4rWsBU5WB372Idz9USilrGXE-253D-26reserved-3D0%26d%3DDwMFaQ%26c%3DOCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag%26r%3DjUfnSyKkfkyVRBIUzlG1GSGGZAZGcznwr8YliSSCjPc%26m%3Dojx3DcDQE2hr2yiIPoSjDlagDXtFEW9J1hoRZUO6cVc%26s%3DiFIxvgyH8sawhze2_wMWmonbVfmpzGFOA8R3KlI9fgk%26e%3D&data=02%7C01%7Comnassar%40UTMB.EDU%7Cdfc37453346e4ab070bd08d5f0e16c7b%7C7bef256d85db4526a72d31aea2546852%7C0%7C0%7C636679773341569223&sdata=ovmkZX%2B6ciRgIIeBp%2FzVWFEPG28gn2iJqlLZZ5w8Kcc%3D&reserved=0>), it seems that it is not possible to properly separate van der Waals and electrostatics calculations using FEP.

However, in contrast to my understanding of the above documuntation and according to this very general response (http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l.2010-2011/1722.htmlww.ks.uiuc.edu-252FResearch-252Fnamd-252Fmailing-5Flist-252Fnamd-2Dl.2010-2D2011-252F1722.html-26data-3D02-257C01-257Comnassar-2540UTMB.EDU-257C099dbcc3f2c0496789c708d5eef97f81-257C7bef256d85db4526a72d31aea2546852-257C0-257C1-257C636677677711073805-26sdata-3DWOaQxK6X-252B038lGSGPbfUnswBbf0vH7-252Bwp-252Ffqfr42-252B74-253D-26reserved-3D0%26d%3DDwMFaQ%26c%3DOCIEmEwdEq_aNlsP4fF3gFqSN-E3mlr2t9JcDdfOZag%26r%3DjUfnSyKkfkyVRBIUzlG1GSGGZAZGcznwr8YliSSCjPc%26m%3Dojx3DcDQE2hr2yiIPoSjDlagDXtFEW9J1hoRZUO6cVc%26s%3D9dY63UuQAwNwG25E8c_uCCkkkQg-oVhpS_0XGIMXKjM%26e%3D&data=02%7C01%7Comnassar%40UTMB.EDU%7Cdfc37453346e4ab070bd08d5f0e16c7b%7C7bef256d85db4526a72d31aea2546852%7C0%7C0%7C636679773341579232&sdata=9MVdvK6kjkvEt2TmdQHc8LK%2BwknzcMlsDGvGc9XE%2BfU%3D&reserved=0>), it is claimed that it is possible to separate vanderwaals and electrostatics.

Below is the input for van der Waals and electrostatics for each lambda value (I am simulating each lambda window independently).

____________________________________________________________
VANDERWAALS CALCULATION/SIMULATION:

# FEP PARAMETERS

source fep.tcl

alch on
alchType FEP
alchFile N3A_ion.fep
alchCol B
alchOutFile $outname.fepout
alchOutFreq 500

alchVdwLambdaEnd 1.0
alchElecLambdaStart 1.0
alchVdWShiftCoeff 5.0
alchDecouple on
alchEquilSteps 1000
set numSteps 6251000

runFEP 0.6875 0.71875 0.03125 $numSteps
______________________________________________________________
ELECTROSTATICS CALCULATION/SIMULATION:

# FEP PARAMETERS

source fep.tcl

alch on
alchType FEP
alchFile N3A_ion.fep
alchCol B
alchOutFile $outname.fepout
alchOutFreq 500

alchVdwLambdaEnd 0.0
alchElecLambdaStart 0.0
alchVdWShiftCoeff 5.0
alchDecouple on
alchEquilSteps 1000
set numSteps 6251000

runFEP 0.6875 0.71875 0.03125 $numSteps

Using the input examples above, am I properly separating electrostatics and van der waals calculations? If so, how is the geometry of the molecule affected when I run the electrostatics calculation? Do I have both NADH and ADP present geometrically, but only the charges are changing?

Your help is invaluable and I really appreciate your response.

Thank you!

Omneya Nassar
PhD Candidate in Pharmacology and Toxicology
University of Texas Medical Branch
301 University Blvd
Galveston, TX 77555
E-mail:
omnassar_at_utmb.edu<mailto:omnassar_at_utmb.edu>
Phone: 409.772.0731

This archive was generated by hypermail 2.1.6 : Thu Dec 05 2019 - 23:20:03 CST