From: Nathan A. Baker <baker@ch...>  20031219 14:14:21

Hi Fabrice  I get this question fairly frequently so I'm going to Cc my reply to the APBS mailing list. There are two ways to get the Coulombic energy of a complex: 1. Use Coulomb's law (apbs/tools/manip/coulomb) and scale by the appropriate dielectric coefficient. 2. Solve Poission's equation with a homogeneous dielectric coefficient to determine the energies of protein A, protein B, and the AB complex. The proteins' atoms must stay at exactly the same place on the grid for each calculation; likewise, the grid spacing, position, etc. cannot change between calculations. The Coulombic energy change for forming the complex is then: E(AB)  E(A)  E(B) Of the two ways, #1 is almost always preferable. Fabrice Leclerc <Fabrice.Leclerc@...> (12192003 00:14:50+0100): >Hi Nathan, >I'm interested in the coulombic energy between 2 molecules. The >calculation using the "coulomb" utility is pretty straightforward. How >do you extactly extract this term from the numerical calculations ? >Thanks. >Fabrice. > > >On Friday, December 12, 2003, at 03:27 PM, Nathan Baker wrote: > > >>Hi Fabrice  >> >>You always need to perform a reference calculation with APBS to >>remove the selfenergies. In other words, if you perform a solvation >>energy calculation, then the uniform dielectric result is your >>"selfenergy". The Coulombic energies can either be obtained from >>numerical (apbs) or analytical (tools/manip/coulomb) manipulations. >> >>Thanks, >> >>Nathan >> >> >>Nathan A. Baker, Assistant Professor >>Washington University in St. Louis >>Dept. of Biochemistry and Molecular Biophysics >>Center for Computational Biology >>700 S. Euclid Ave., Campus Box 8036, St. Louis, MO 63110 >>Phone: (314) 3622040, Fax: (314) 3620234 >>URL: http://www.biochem.wustl.edu/~baker/ >> >> >>Original Message >>From: Fabrice Leclerc [mailto:Fabrice.Leclerc@...] >>Sent: Friday, December 12, 2003 6:42 AM >>To: baker@... >>Subject: Re: convert uhbd input to apbs >> >>Hi Nathan, >>thanks for your quick reply. >>I'm wondering whether it's possible to decompose the total >>electrostatic energy into its different contributions rather than by >>peratom components with the calcenergy flag. I mean to have the >>electrostatic coulombic and self energies as well. >> >>On Thursday, December 11, 2003, at 02:47 PM, Nathan Baker wrote: >> >>Hello  >> >>While you cannot convert UHBD input to APBS in a straightforward way >>(there are internal routines...), you can convert APBS to UHBD with >>the utility in the apbs/tools directory. That should do what you >>need. >> >>For the parallel focusing calculations, you should choose your fine >>and coarse grid dimensions/spacings to match what you would normally >>give to UHBD. Based on the list you gave below, I'd suggest >> >>1. dime 65 65 65, glen 128 128 128 >>2. dime 97 97 97, glen 97 97 97 >>3. dime 97 97 161, glen 48 48 80 >> >>This will preserve the grid spacing (important for comparisons) but >>will result in slightly larger calculation domains. >> >>Thanks, >> >>Nathan >> >> >>Nathan A. Baker, Assistant Professor >>Washington University in St. Louis >>Dept. of Biochemistry and Molecular Biophysics >>Center for Computational Biology >>700 S. Euclid Ave., Campus Box 8036, St. Louis, MO 63110 >>Phone: (314) 3622040, Fax: (314) 3620234 >>URL: http://www.biochem.wustl.edu/~baker/ >> >> >>Original Message >>From: Fabrice Leclerc [mailto:Fabrice.Leclerc@...] >>Sent: Wednesday, December 10, 2003 5:59 PM >>To: baker@... >>Subject: convert uhbd input to apbs >> >>Hi Nathan, >>I'm doing some tests with APBS (0.2.6) and want to compare with some >>previous calculations I've done with UHBD. Is there any way to easily >>convert a UHBD input file into a APBS input file ? >>In the parallel version, it seems to be necessary to specify cglen >>and fglen but it not obvious to me what values I should use to >>reproduce a setting similar to what I had with UHBD. In the UHBD >>calculation, I use to have 3 calculations (2 focussing) with >>different grid sizes: >> >>1. spacing 2.0 dime 51 56 65 >>2. spacing 1.0 dime 61 70 88 >>3. spacing 0.5 dime 83 100 137 >> >>I have to say I couldn't figure out how to get a similar setting with >>APBS. Can you give me some hint ? Thanks. >>Fabrice. >> >> >>\ /\ >>\/ \______________________________________ >>/\ >>AU Fabrice Leclerc, Ph.D. >>() Laboratoire de Maturation des ARN >>GC et Enzymologie Moléculaire (MAEM) >>\/ UMR 7567 CNRSUHP Nancy I >>/\ Faculté des Sciences, B.P. 239 >>UA 54506 VandoeuvrelèsNancy Cedex >>() Tel. (33) 3 83 68 43 17 >>CG Fax. (33) 3 83 68 43 07 >>\/ Fabrice.Leclerc@... >>/\ _____________________________________ >>/ \ / >>\/ >> >> > > End of message from Fabrice Leclerc.  Nathan A. Baker, Assistant Professor Washington University in St. Louis School of Medicine Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology 700 S. Euclid Ave., Campus Box 8036, St. Louis, MO 63110 Phone: (314) 3622040, Fax: (314) 3620234 URL: http://www.biochem.wustl.edu/~baker PGP key: http://cholla.wustl.edu/~baker/pubkey.asc 
From: Fabrice Leclerc <Fabrice.L<eclerc@ma...>  20040115 09:07:25

Hi, I'm doing some comparison between UHBD and APBS on a series of test calculation. I have a similar setup for both programs on a bimolecular complex (grid spacing, solute and solvent dielectric, ionic strength, temperature, etc) . I have calculated the coulombic energy of interaction between the 2 molecules of the complex using UHBD and APBS. The calculations performed as described below (message from december 19) give similar results between UHBD and APBS. I'm interested now in the total electrostatic energy term for the complex ("print elec energy all end" in UHBD). I don't get equivalent results between the 2 programs. Is this just due to the difference in grid mapping between the 2 programs ? Is there any correspondence between the numerical values given by the 2 programs as the total electrostatic energy ? If I want to calculate the the electrostatic energy to transfer the complex from the gas phase (dielectric = 1) to the aqueous phase (dielectric 78), it can be done with UHBD in a single calculation ("print elec tenerg all end") after printing and subtracting the self energy ("print elec self all end") and the coulombic grid contribution ("print elec fdcoul all end"). What is the correct way to do this calculation with APBS ? On Friday, December 19, 2003, at 01:14 PM, Nathan A. Baker wrote: > Hi Fabrice  > > I get this question fairly frequently so I'm going to Cc my reply to > the APBS mailing list. > > There are two ways to get the Coulombic energy of a complex: > 1. Use Coulomb's law (apbs/tools/manip/coulomb) and scale by the > appropriate dielectric coefficient. > 2. Solve Poission's equation with a homogeneous dielectric > coefficient to determine the energies of protein A, protein B, and > the > AB complex. The proteins' atoms must stay at exactly the same > place on the grid for each calculation; likewise, the grid > spacing, position, etc. cannot change between calculations. The > Coulombic energy change for forming the complex is then: > E(AB)  E(A)  E(B) 
From: Nathan A. Baker <baker@ch...>  20040115 14:32:59

Hello  >I'm doing some comparison between UHBD and APBS on a series of test >calculation. I have a similar setup for both programs on a bimolecular >complex (grid spacing, solute and solvent dielectric, ionic strength, >temperature, etc) . I have calculated the coulombic energy of >interaction between the 2 molecules of the complex using UHBD and APBS. >The calculations performed as described below (message from december >19) give similar results between UHBD and APBS. Great! >I'm interested now in the total electrostatic energy term for the >complex ("print elec energy all end" in UHBD). I don't get >equivalent results between the 2 programs. Is this just due to the >difference in grid mapping between the 2 programs ? Is there any >correspondence between the numerical values given by the 2 programs >as the total electrostatic energy ? The total electrostatic energy is not meaningful for a variety of reasons, including: charge and surface discretizations, etc. Therefore; you should always compare a difference of the total energy with respect to a reference state. >If I want to calculate the the electrostatic energy to transfer the >complex from the gas phase (dielectric = 1) to the aqueous phase >(dielectric 78), it can be done with UHBD in a single calculation >("print elec tenerg all end") after printing and subtracting the self >energy ("print elec self all end") and the coulombic grid >contribution ("print elec fdcoul all end"). What is the correct way >to do this calculation with APBS ? UHBD uses a "trick" to compute the uniform dielectric energy semianalytically (see Luty et al, J Comp Chem 13 (6) 768771, 1992) while APBS requires two calculations: one for the solvated form and another for the reference (uniform dielectric) system. Thanks, Nathan >On Friday, December 19, 2003, at 01:14 PM, Nathan A. Baker wrote: > >>Hi Fabrice  >> >>I get this question fairly frequently so I'm going to Cc my reply to >>the APBS mailing list. >> >>There are two ways to get the Coulombic energy of a complex: >>1. Use Coulomb's law (apbs/tools/manip/coulomb) and scale by the >> appropriate dielectric coefficient. >>2. Solve Poission's equation with a homogeneous dielectric >> coefficient to determine the energies of protein A, protein B, and >>the >> AB complex. The proteins' atoms must stay at exactly the same >> place on the grid for each calculation; likewise, the grid >> spacing, position, etc. cannot change between calculations. The >> Coulombic energy change for forming the complex is then: >> E(AB)  E(A)  E(B) > > >_______________________________________________ >apbsusers mailing list >apbsusers@... >http://cholla.wustl.edu/mailman/listinfo/apbsusers End of message from Fabrice Leclerc.  Nathan A. Baker, Assistant Professor Washington University in St. Louis School of Medicine Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology 700 S. Euclid Ave., Campus Box 8036, St. Louis, MO 63110 Phone: (314) 3622040, Fax: (314) 3620234 URL: http://www.biochem.wustl.edu/~baker PGP key: http://cholla.wustl.edu/~baker/pubkey.asc 
From: Nathan A. Baker <baker@ch...>  20040115 14:34:19

I forgot to mention: you should look at apbs/examples/born and apbs/examples/solv for examples of how to compute solvation energies in APBS. Thanks, Nathan Nathan A. Baker <baker@...> (01152004 06:32:530600): >Hello  > >>I'm doing some comparison between UHBD and APBS on a series of test >>calculation. I have a similar setup for both programs on a bimolecular >>complex (grid spacing, solute and solvent dielectric, ionic strength, >>temperature, etc) . I have calculated the coulombic energy of >>interaction between the 2 molecules of the complex using UHBD and APBS. >>The calculations performed as described below (message from december >>19) give similar results between UHBD and APBS. > >Great! > >>I'm interested now in the total electrostatic energy term for the >>complex ("print elec energy all end" in UHBD). I don't get >>equivalent results between the 2 programs. Is this just due to the >>difference in grid mapping between the 2 programs ? Is there any >>correspondence between the numerical values given by the 2 programs >>as the total electrostatic energy ? > >The total electrostatic energy is not meaningful for a variety of >reasons, including: charge and surface discretizations, etc. >Therefore; you should always compare a difference of the total energy >with respect to a reference state. > >>If I want to calculate the the electrostatic energy to transfer the >>complex from the gas phase (dielectric = 1) to the aqueous phase >>(dielectric 78), it can be done with UHBD in a single calculation >>("print elec tenerg all end") after printing and subtracting the self >>energy ("print elec self all end") and the coulombic grid >>contribution ("print elec fdcoul all end"). What is the correct way >>to do this calculation with APBS ? > >UHBD uses a "trick" to compute the uniform dielectric energy >semianalytically (see Luty et al, J Comp Chem 13 (6) 768771, 1992) >while APBS requires two calculations: one for the solvated form and >another for the reference (uniform dielectric) system. > >Thanks, > >Nathan > >>On Friday, December 19, 2003, at 01:14 PM, Nathan A. Baker wrote: >> >>>Hi Fabrice  >>> >>>I get this question fairly frequently so I'm going to Cc my reply to >>>the APBS mailing list. >>> >>>There are two ways to get the Coulombic energy of a complex: >>>1. Use Coulomb's law (apbs/tools/manip/coulomb) and scale by the >>> appropriate dielectric coefficient. >>>2. Solve Poission's equation with a homogeneous dielectric >>> coefficient to determine the energies of protein A, protein B, and >>>the >>> AB complex. The proteins' atoms must stay at exactly the same >>> place on the grid for each calculation; likewise, the grid >>> spacing, position, etc. cannot change between calculations. The >>> Coulombic energy change for forming the complex is then: >>> E(AB)  E(A)  E(B) >> >> >>_______________________________________________ >>apbsusers mailing list >>apbsusers@... >>http://cholla.wustl.edu/mailman/listinfo/apbsusers >End of message from Fabrice Leclerc. > > >Nathan A. Baker, Assistant Professor >Washington University in St. Louis School of Medicine >Dept. of Biochemistry and Molecular Biophysics >Center for Computational Biology >700 S. Euclid Ave., Campus Box 8036, St. Louis, MO 63110 >Phone: (314) 3622040, Fax: (314) 3620234 >URL: http://www.biochem.wustl.edu/~baker >PGP key: http://cholla.wustl.edu/~baker/pubkey.asc End of message from Nathan A. Baker.  Nathan A. Baker, Assistant Professor Washington University in St. Louis School of Medicine Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology 700 S. Euclid Ave., Campus Box 8036, St. Louis, MO 63110 Phone: (314) 3622040, Fax: (314) 3620234 URL: http://www.biochem.wustl.edu/~baker PGP key: http://cholla.wustl.edu/~baker/pubkey.asc 
From: Fabrice Leclerc <Fabrice.L<eclerc@ma...>  20040116 13:28:51

Hi, to follow up my previous questions on the comparison between UHBD and APBS, I've done a calculation of the electrostatic free energy change for transfer of a DNA decamer between the gas phase (eps=1) and the aqueous phase (eps=78). As mentionned by Nathan in his reply, UHBD does "compute the uniform dielectric energy semianalytically" and gives directly the global result for the energy change: Delta E = E(eps[solute]=1,eps[solvent=78.0])  E(eps[solute]=1,eps[solvent]=1), while both calculations have to be explicitly set up to get the two terms from the equation with APBS. The testcases is taken from the examples given with the UHBD code (version 5.1, examples/dna.inp). The electrostatic free energy change associated with this process, calculated by UHBD, is 4526.6 kcal/mol, e.g. 18921 kJ/mol. I have converted the UHBD input into an APBS input using the same parameters (grid spacing, temparature, and so on) as much as possible (both input and output files are attached in case you may want to check). The numerical value given by APBS is 3665.64 kJ/mol. Why are these 2 calculations not equivalent ? On Thursday, January 15, 2004, at 01:34 PM, Nathan A. Baker wrote: > I forgot to mention: you should look at apbs/examples/born and > apbs/examples/solv for examples of how to compute solvation energies > in APBS. > > Thanks, > > Nathan > >>> If I want to calculate the the electrostatic energy to transfer the >>> complex from the gas phase (dielectric = 1) to the aqueous phase >>> (dielectric 78), it can be done with UHBD in a single calculation >>> ("print elec tenerg all end") after printing and subtracting the self >>> energy ("print elec self all end") and the coulombic grid >>> contribution ("print elec fdcoul all end"). What is the correct way >>> to do this calculation with APBS ? >> >> UHBD uses a "trick" to compute the uniform dielectric energy >> semianalytically (see Luty et al, J Comp Chem 13 (6) 768771, 1992) >> while APBS requires two calculations: one for the solvated form and >> another for the reference (uniform dielectric) system. >> >> Thanks, >> >> Nathan >> 
From: Nathan Baker <baker@bi...>  20040117 05:52:30

I'm not sure  I'll have to look into this some more. However, my initial guess that neither calculation is converged (with respect to grid spacing/placement) and you're just seeing different, poorly converged results. This could be particularly true with APBS which uses a cruder surface definition than implemented in APBS by Gilson and other folks. In any case, I'll check this problem out and see what I can find. Thanks, Nathan  Nathan A. Baker, Assistant Professor Washington University in St. Louis School of Medicine Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology 700 S. Euclid Ave., Campus Box 8036, St. Louis, MO 63110 Phone: (314) 3622040, Fax: (314) 3620234 URL: http://www.biochem.wustl.edu/~baker PGP key: http://cholla.wustl.edu/~baker/pubkey.asc > Original Message > From: Fabrice Leclerc [mailto:Fabrice.Leclerc@...] > Sent: Friday, January 16, 2004 5:36 AM > To: Nathan A. Baker > Cc: apbsusers@... > Subject: Re: [Apbsusers] Re: convert uhbd input to apbs > > > Hi, > to follow up my previous questions on the comparison between UHBD and > APBS, I've done a calculation of the electrostatic free energy change > for transfer of a DNA decamer between the gas phase (eps=1) and the > aqueous phase (eps=78). As mentionned by Nathan in his reply, > UHBD does > "compute the uniform dielectric energy semianalytically" and gives > directly the global result for the energy change: > Delta E = E(eps[solute]=1,eps[solvent=78.0])  > E(eps[solute]=1,eps[solvent]=1), > while both calculations have to be explicitly set up to get the two > terms from the equation with APBS. > The testcases is taken from the examples given with the UHBD code > (version 5.1, examples/dna.inp). The electrostatic free energy change > associated with this process, calculated by UHBD, is 4526.6 > kcal/mol, > e.g. 18921 kJ/mol. > I have converted the UHBD input into an APBS input using the same > parameters (grid spacing, temparature, and so on) as much as possible > (both input and output files are attached in case you may want to > check). The numerical value given by APBS is 3665.64 kJ/mol. > Why are these 2 calculations not equivalent ? > > > On Thursday, January 15, 2004, at 01:34 PM, Nathan A. Baker wrote: > > > I forgot to mention: you should look at apbs/examples/born and > > apbs/examples/solv for examples of how to compute solvation > energies > > in APBS. > > > > Thanks, > > > > Nathan > > > >>> If I want to calculate the the electrostatic energy to > transfer the > >>> complex from the gas phase (dielectric = 1) to the aqueous phase > >>> (dielectric 78), it can be done with UHBD in a single calculation > >>> ("print elec tenerg all end") after printing and > subtracting the self > >>> energy ("print elec self all end") and the coulombic grid > >>> contribution ("print elec fdcoul all end"). What is the > correct way > >>> to do this calculation with APBS ? > >> > >> UHBD uses a "trick" to compute the uniform dielectric energy > >> semianalytically (see Luty et al, J Comp Chem 13 (6) > 768771, 1992) > >> while APBS requires two calculations: one for the > solvated form and > >> another for the reference (uniform dielectric) system. > >> > >> Thanks, > >> > >> Nathan > >> > > > 