Work at SourceForge, help us to make it a better place! We have an immediate need for a Support Technician in our San Francisco or Denver office.
Close
From: David Dynerman <dynerman@cs...>  20080225 23:22:23

Hello, I'm trying to use APBS to calculate the association energy for Trypsin/Pancreatic Trypsin Inhibitor (2PTC) The experimental result of 18.1 kcal/mol is from: "Trypsinpancreatic trypsin inhibitor association. Dynamics of the interaction and role of disulfide bridges" Jean Pierre Vincent and Michel Lazdunski Biochemistry 1972 Vol 11 Iss 16 pp29672977 These are the steps I'm taking: 1) Download 2PTC from the PDB 2) Strip all nonatom records from the PDB 3) Save the E chain as 2PTC_r.pdb 4) Save the I chain as 2PTC_l.pdb 5) Run pdb2pqr apbsinput ff=amber on the 3 PDB files 6) Concatenate the generated APBS output files to that all three are processed 7) Run APBS, output: print energy complexsolv  ligandsolv  receptorsolv end where complexsolv is due to 2PTC.pqr, ligandsolv is due to 2PTC_l.pqr and receptorsolv is due to 2PTC_r.pqr I'm not deeply familiar with APBS (only recently started using it) so most of the APBS parameters are the ones pdb2pqr generated. I reused the complex's cglen fglen cgcent and fgcent for all three electrostatic evaluations. I must be doing something wrong, since the energies I get are of the wrong sign and magnitude (2.061272956069E+02 kJ/mol) If I subtract off the coulombic interactions: (columb 2PTC.pqr  coulomb 2PTC_l.pqr  coulomb 2PTC_r.pqr)/2 I end up with 61 kcal/mol, but I'm not sure if that's still measuring the association energy. The questions I have are the following: Is my interpretation of this calculation as the binding association correct? Is it correct to subtract the coulombic energy from the APBS energy to get the solvent association energy? Are there any easy things to check (other than grid considerations, which I think I have uniform) to try and get the APBS value closer to the experimental value? Thank you, David 
From: Nathan Baker <baker@cc...>  20080226 17:45:43

Hi David  As discussed on the list previously, the polar contribution to proteinprotein binding energies (particularly when calculated with PB methods and certain discontinuous surface definitions) is not always favorable and is often offset by changes in nonpolar solvation energy. Additionally, it can be difficult to converge free energy changes with respect to grid spacing using the method you describe. Therefore, I suggest the following tests: (1) Check the sensitivity of your calculations to the protein dielectric (e.g., values between 212)  see work by Elcock, Sept, and McCammon for a discussion of these values. (2) Check the sensitivity of your calculations to the surface definition, including van der Waals (srad 0.0)  see work by Dong and Zhou for a discussion of surface sensitivity (3) Check the sensitivity of your calculations to the number of grid points  does increasing the number of grid points (and thereby increasing the mesh resolution) change your answer significantly? If so, the results are not converged. You might also try performing your binding free energy calculations a slightly different manner following the apbs/examples/actindimer example. This method is often a bit less sensitive to grid spacing. Finally, you should keep in mind that implicit solvent calculations are generally used for relative free energy calculations: e.g., mutagenesis effects, salt dependence, protonation effects, etc. You may be able to pose your problem in terms of relative binding free energies  what are you interested in learning about this system?  Nathan On Mon, Feb 25, 2008 at 5:24 PM, David Dynerman <dynerman@...> wrote: > Hello, > > I'm trying to use APBS to calculate the association energy for > Trypsin/Pancreatic Trypsin Inhibitor (2PTC) > > The experimental result of 18.1 kcal/mol is from: > > "Trypsinpancreatic trypsin inhibitor association. Dynamics of the > interaction and role of disulfide bridges" > Jean Pierre Vincent and Michel Lazdunski Biochemistry 1972 Vol 11 Iss 16 > pp29672977 > > These are the steps I'm taking: > > 1) Download 2PTC from the PDB > 2) Strip all nonatom records from the PDB > 3) Save the E chain as 2PTC_r.pdb > 4) Save the I chain as 2PTC_l.pdb > 5) Run pdb2pqr apbsinput ff=amber on the 3 PDB files > 6) Concatenate the generated APBS output files to that all three are > processed > 7) Run APBS, output: > > print energy complexsolv  ligandsolv  receptorsolv end > > where complexsolv is due to 2PTC.pqr, ligandsolv is due to 2PTC_l.pqr > and receptorsolv is due to 2PTC_r.pqr > > I'm not deeply familiar with APBS (only recently started using it) so > most of the APBS parameters are the ones pdb2pqr generated. I reused > the complex's cglen fglen cgcent and fgcent for all three electrostatic > evaluations. > > I must be doing something wrong, since the energies I get are of the > wrong sign and magnitude (2.061272956069E+02 kJ/mol) > > If I subtract off the coulombic interactions: (columb 2PTC.pqr  coulomb > 2PTC_l.pqr  coulomb 2PTC_r.pqr)/2 I end up with 61 kcal/mol, but I'm > not sure if that's still measuring the association energy. > > The questions I have are the following: > > Is my interpretation of this calculation as the binding association > correct? Is it correct to subtract the coulombic energy from the APBS > energy to get the solvent association energy? > > Are there any easy things to check (other than grid considerations, > which I think I have uniform) to try and get the APBS value closer to > the experimental value? > > Thank you, > David > >  > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > apbsusers mailing list > apbsusers@... > https://lists.sourceforge.net/lists/listinfo/apbsusers >  Associate Professor, Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology, Washington University in St. Louis Web: http://cholla.wustl.edu/ 
From: David Dynerman <dynerman@cs...>  20080226 21:54:28

Nathan, Thank you very much for the references and tips  I'll give them a try later today and see where they get me. I should preface everything by mentioning that my background is in computation  I've only recently gotten into computational biology stuff, so any references and clarifications are greatly appreciated. My overall goal is analyzing the accuracy of a desolvation energy algorithm. To that end, I hope to use APBS to compute desolvation energies and use them as a "gold standard" to compare against. As I understand it, given the following thermodynamic cycle: (L = ligand, R = receptor, C = complex) Warning: ASCII diagram ahead 1 L + R (vac) > C (vac)     2 3 4 V V V L + R (solv) > C (solv) 5 the desolvation energy will be 5  1. As the first step on this road, I'm trying to calculate 5 (e.g. complexsolv  ligandsolv  receptorsolv) I also believe (correct me if I'm wrong) that, in theory at least, 5 should correspond to experimentally determined association energies. You mentioned that there may be a way to achieve the same calculation with less numericalwoes using relative energies  if there's a better way of doing this, I would greatly appreciate any info you'd be willing to provide. Thanks, David Nathan Baker wrote: > Hi David  > > As discussed on the list previously, the polar contribution to > proteinprotein binding energies (particularly when calculated with PB > methods and certain discontinuous surface definitions) is not always > favorable and is often offset by changes in nonpolar solvation energy. > Additionally, it can be difficult to converge free energy changes > with respect to grid spacing using the method you describe. > Therefore, I suggest the following tests: > > (1) Check the sensitivity of your calculations to the protein > dielectric (e.g., values between 212)  see work by Elcock, Sept, > and McCammon for a discussion of these values. > (2) Check the sensitivity of your calculations to the surface > definition, including van der Waals (srad 0.0)  see work by Dong and > Zhou for a discussion of surface sensitivity > (3) Check the sensitivity of your calculations to the number of grid > points  does increasing the number of grid points (and thereby > increasing the mesh resolution) change your answer significantly? If > so, the results are not converged. You might also try performing your > binding free energy calculations a slightly different manner following > the apbs/examples/actindimer example. This method is often a bit > less sensitive to grid spacing. > > Finally, you should keep in mind that implicit solvent calculations > are generally used for relative free energy calculations: e.g., > mutagenesis effects, salt dependence, protonation effects, etc. You > may be able to pose your problem in terms of relative binding free > energies  what are you interested in learning about this system? > >  Nathan > > > > On Mon, Feb 25, 2008 at 5:24 PM, David Dynerman <dynerman@...> wrote: >> Hello, >> >> I'm trying to use APBS to calculate the association energy for >> Trypsin/Pancreatic Trypsin Inhibitor (2PTC) >> >> The experimental result of 18.1 kcal/mol is from: >> >> "Trypsinpancreatic trypsin inhibitor association. Dynamics of the >> interaction and role of disulfide bridges" >> Jean Pierre Vincent and Michel Lazdunski Biochemistry 1972 Vol 11 Iss 16 >> pp29672977 >> >> These are the steps I'm taking: >> >> 1) Download 2PTC from the PDB >> 2) Strip all nonatom records from the PDB >> 3) Save the E chain as 2PTC_r.pdb >> 4) Save the I chain as 2PTC_l.pdb >> 5) Run pdb2pqr apbsinput ff=amber on the 3 PDB files >> 6) Concatenate the generated APBS output files to that all three are >> processed >> 7) Run APBS, output: >> >> print energy complexsolv  ligandsolv  receptorsolv end >> >> where complexsolv is due to 2PTC.pqr, ligandsolv is due to 2PTC_l.pqr >> and receptorsolv is due to 2PTC_r.pqr >> >> I'm not deeply familiar with APBS (only recently started using it) so >> most of the APBS parameters are the ones pdb2pqr generated. I reused >> the complex's cglen fglen cgcent and fgcent for all three electrostatic >> evaluations. >> >> I must be doing something wrong, since the energies I get are of the >> wrong sign and magnitude (2.061272956069E+02 kJ/mol) >> >> If I subtract off the coulombic interactions: (columb 2PTC.pqr  coulomb >> 2PTC_l.pqr  coulomb 2PTC_r.pqr)/2 I end up with 61 kcal/mol, but I'm >> not sure if that's still measuring the association energy. >> >> The questions I have are the following: >> >> Is my interpretation of this calculation as the binding association >> correct? Is it correct to subtract the coulombic energy from the APBS >> energy to get the solvent association energy? >> >> Are there any easy things to check (other than grid considerations, >> which I think I have uniform) to try and get the APBS value closer to >> the experimental value? >> >> Thank you, >> David >> >>  >> This SF.net email is sponsored by: Microsoft >> Defy all challenges. Microsoft(R) Visual Studio 2008. >> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ >> _______________________________________________ >> apbsusers mailing list >> apbsusers@... >> https://lists.sourceforge.net/lists/listinfo/apbsusers >> > > > 
From: Nathan Baker <baker@cc...>  20080227 13:48:29

Hi David  > My overall goal is analyzing the accuracy of a desolvation energy > algorithm. To that end, I hope to use APBS to compute desolvation > energies and use them as a "gold standard" to compare against. As I > understand it, given the following thermodynamic cycle: (L = ligand, > R = receptor, C = complex) > > Warning: ASCII diagram ahead > > 1 > L + R (vac) > C (vac) >    >  2 3 4 > V V V > L + R (solv) > C (solv) > 5 > > the desolvation energy will be 5  1. Yes, this is correct. > As the first step on this road, I'm trying to calculate 5 (e.g. > complexsolv  ligandsolv  receptorsolv) OK  both polar and nonpolar components? > I also believe (correct me if I'm wrong) that, in theory at least, 5 > should correspond to experimentally determined association energies. Only if 5 includes polar and nonpolar solvation energy changes, molecular mechanical changes between the bound and unbound states, and entropic terms. > You mentioned that there may be a way to achieve the same > calculation with less numericalwoes using relative energies  if > there's a better way of doing this, I would greatly appreciate any > info you'd be willing to provide. Yes, the apbs/examples/actindimer example computes 5 directly for polar solvation energy change upon the binding of *rigid* actin dimers. Hope this helps!  Nathan > > Thanks, > David > > Nathan Baker wrote: >> Hi David  >> As discussed on the list previously, the polar contribution to >> proteinprotein binding energies (particularly when calculated with >> PB >> methods and certain discontinuous surface definitions) is not always >> favorable and is often offset by changes in nonpolar solvation >> energy. >> Additionally, it can be difficult to converge free energy changes >> with respect to grid spacing using the method you describe. >> Therefore, I suggest the following tests: >> (1) Check the sensitivity of your calculations to the protein >> dielectric (e.g., values between 212)  see work by Elcock, Sept, >> and McCammon for a discussion of these values. >> (2) Check the sensitivity of your calculations to the surface >> definition, including van der Waals (srad 0.0)  see work by Dong >> and >> Zhou for a discussion of surface sensitivity >> (3) Check the sensitivity of your calculations to the number of grid >> points  does increasing the number of grid points (and thereby >> increasing the mesh resolution) change your answer significantly? If >> so, the results are not converged. You might also try performing >> your >> binding free energy calculations a slightly different manner >> following >> the apbs/examples/actindimer example. This method is often a bit >> less sensitive to grid spacing. >> Finally, you should keep in mind that implicit solvent calculations >> are generally used for relative free energy calculations: e.g., >> mutagenesis effects, salt dependence, protonation effects, etc. You >> may be able to pose your problem in terms of relative binding free >> energies  what are you interested in learning about this system? >>  Nathan >> On Mon, Feb 25, 2008 at 5:24 PM, David Dynerman >> <dynerman@...> wrote: >>> Hello, >>> >>> I'm trying to use APBS to calculate the association energy for >>> Trypsin/Pancreatic Trypsin Inhibitor (2PTC) >>> >>> The experimental result of 18.1 kcal/mol is from: >>> >>> "Trypsinpancreatic trypsin inhibitor association. Dynamics of the >>> interaction and role of disulfide bridges" >>> Jean Pierre Vincent and Michel Lazdunski Biochemistry 1972 Vol 11 >>> Iss 16 >>> pp29672977 >>> >>> These are the steps I'm taking: >>> >>> 1) Download 2PTC from the PDB >>> 2) Strip all nonatom records from the PDB >>> 3) Save the E chain as 2PTC_r.pdb >>> 4) Save the I chain as 2PTC_l.pdb >>> 5) Run pdb2pqr apbsinput ff=amber on the 3 PDB files >>> 6) Concatenate the generated APBS output files to that all three are >>> processed >>> 7) Run APBS, output: >>> >>> print energy complexsolv  ligandsolv  receptorsolv end >>> >>> where complexsolv is due to 2PTC.pqr, ligandsolv is due to >>> 2PTC_l.pqr >>> and receptorsolv is due to 2PTC_r.pqr >>> >>> I'm not deeply familiar with APBS (only recently started using it) >>> so >>> most of the APBS parameters are the ones pdb2pqr generated. I re >>> used >>> the complex's cglen fglen cgcent and fgcent for all three >>> electrostatic >>> evaluations. >>> >>> I must be doing something wrong, since the energies I get are of the >>> wrong sign and magnitude (2.061272956069E+02 kJ/mol) >>> >>> If I subtract off the coulombic interactions: (columb 2PTC.pqr  >>> coulomb >>> 2PTC_l.pqr  coulomb 2PTC_r.pqr)/2 I end up with 61 kcal/mol, but >>> I'm >>> not sure if that's still measuring the association energy. >>> >>> The questions I have are the following: >>> >>> Is my interpretation of this calculation as the binding association >>> correct? Is it correct to subtract the coulombic energy from the >>> APBS >>> energy to get the solvent association energy? >>> >>> Are there any easy things to check (other than grid considerations, >>> which I think I have uniform) to try and get the APBS value closer >>> to >>> the experimental value? >>> >>> Thank you, >>> David >>> >>>  >>> This SF.net email is sponsored by: Microsoft >>> Defy all challenges. Microsoft(R) Visual Studio 2008. >>> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ >>> _______________________________________________ >>> apbsusers mailing list >>> apbsusers@... >>> https://lists.sourceforge.net/lists/listinfo/apbsusers >>> >  Associate Professor, Dept. of Biochemistry and Molecular Biophysics Center for Computational Biology, Washington University in St. Louis Web: http://cholla.wustl.edu/ 