From owner-chemistry@ccl.net Tue May 24 05:57:00 2011 From: "JL jeffrey.lebowski{:}gmx.de" To: CCL Subject: CCL:G: AW: G: Free energy of solvation Message-Id: <-44736-110524045937-1296-jaEmQoKaxaNeptGDzKHjhA[*]server.ccl.net> X-Original-From: "JL" Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset="iso-8859-1" Date: Tue, 24 May 2011 10:59:25 +0200 MIME-Version: 1.0 Sent to CCL by: "JL" [jeffrey.lebowski|gmx.de] Thanks for the quick reply!!! OK, UAHF was recommended in the G03 manual that is why I chose them although I did KS-DFT calculations. I'll try also with UAKS. What I'm looking for is the Delta G between to Isomers A and B in solution. I have read that this is the difference between G(A,gas)+G(A,solv) and G(B,gas)+G(B,solv) where G(Gas) is the gas phase free energy and G(solv) is the free energy of solvation which one has to add to G(gas). If I understood you correct the G(solv) does not contain any contribution > from rotational and translational degrees of freedom. But isn't that actually right as those degrees of freedom are not present in solution? I read that to a very good approximation there is no rotation and translation in solution only vibrations. Is that right, a little right, completely wrong? If it is right than shouldn't one take for the G(Gas) also only the contributions from vibrations? That would give: DeltaG (in solutuion) = (G(B,gas_vib) + G(B,solv)) - (G(A,gas_vib) + G(A,solv)) Is that complete nonsense? Sorry for those possibly "stupid" questions/ideas? Thanks again in advance for any help!!! Cheers! -----Ursprüngliche Nachricht----- Von: owner-chemistry+jeffrey.lebowski==gmx.de[*]ccl.net [mailto:owner-chemistry+jeffrey.lebowski==gmx.de[*]ccl.net] Im Auftrag von ierre Archirel pierre.archirel%a%u-psud.fr Gesendet: Montag, 23. Mai 2011 17:53 An: Doe, John Betreff: CCL:G: Free energy of solvation Sent to CCL by: " ierre Archirel" [pierre.archirel() u-psud.fr] This is an answer to John. You are rising fondamental issues of the reaction field (RF) methods. In a few points: 1- If you consider the solubility of a molecule in the gas phase, the right value of the Standard Free Energy of solvation reads: DeltaG^0_solv=DeltaG^RF_solv + RTLn 24.5 (i.e. +0.08 eV) taking account of the different conventions of standard states in gas phase (1atm, ie 1/24.5M) and in solution (1M). You will find this in the Cramer book (Computational Chemistry, Wiley 2004 p. 378). I am afraid this will make your calculated value still more positive. 2- the value of Delta G is very sensitive to the values of the radii you are using. I would say that using DFT you should use radii=uaks. 3- the translation-rotation is an awful question. Radii values such as UAHF and UAKS have been optimised, so as to reproduce experimental values of the DeltaG_solv of a list of molecules, mainly in water. The RF Delta G is an electrostatic Delta G, that is the RF Delta S gives the entropy variation of the solvent when the solute is present. The translation-rotation entropy of the solute is certainly absent from the RF Delta G_solv. 4- so you are tied up to the fact that the radii have been optimised to yield "good" values with a false formula (without translation-rotation entropy) 5- you may try to add it by your own, taking for the gas phase the usual values (Sackur-Tetrode and rigid rotor) and for the condensed phase some approximated values (look at M. Amzel and X. Siebert papers). 6- if you manage to do it you will introduce the decrease of the translation-rotation entropy when your molecule is dissolved. I am afraid this will give you a DeltaG_solv more positive, again... This answer is too long, let's wait for other comments! (by the way if you want to try the SDD radii of gaussian09, you may send me your molecule. SMD cavitation effects are small) Pierre Archirel LCP, Universite Paris-Sud France pierre.archirel%u-psud.frhttp://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txt From owner-chemistry@ccl.net Tue May 24 07:44:00 2011 From: "Pierre Archirel pierre.archirel- -u-psud.fr" To: CCL Subject: CCL: Free energy of solvation Message-Id: <-44737-110524074222-22783-EyKEL0Uy92B9fDZaKfwoew:server.ccl.net> X-Original-From: "Pierre Archirel" Date: Tue, 24 May 2011 07:42:20 -0400 Sent to CCL by: "Pierre Archirel" [pierre.archirel^-^u-psud.fr] Dear John, If you are calculating a difference the question is much simpler because entropy effects cancel out. Anyway: 1- Considering that translation and rotation are simply quenched in solution is rather true for energies but completely wrong for entropies. In the vacuum the Sackur-Tetrode and rigid rotor formulas show that the TS contributions amount to (roughly) 0.4 eV (translation) and 0.4 eV (rotation) for a middle size molecule (these values only depend on the log of the mass), 0.8 eV in total therefore. In a condensed phase these contributions may take all the values between zero and the vacuum values. Zero is a plausible value in solids, I consider that in water 30% of the vacuum values is plausible, this amounts to .25 eV, this is not small! 2- But you are looking for a difference! Your two isomers have the same mass and presumably inertial moments, close to each other. So the (large) entropy contributions from translation and rotation nicely cancel out and your formula DeltaG (in solution) = (G(B,gas_vib) + DeltaG(B,solv)) - (G(A,gas_vib) + DeltaG(A,solv)) eventually quite correct. 3- as for vibrations their effect can be easily evaluated through harmonic analysis in the vacuum. 4- I emphasize that translation and rotation entropies are not small in usual solvents. They are considered zero by people who do not know how to calculate them. Pierre Archirel LCP, Universite Paris-Sud France pierre.archirel|u-psud.fr From owner-chemistry@ccl.net Tue May 24 09:04:01 2011 From: "Andreas Klamt klamt- -cosmologic.de" To: CCL Subject: CCL: Free energy of solvation Message-Id: <-44738-110524090212-13768-uKxuO21mT4w7naKs8bEBYw(!)server.ccl.net> X-Original-From: Andreas Klamt Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=ISO-8859-15; format=flowed Date: Tue, 24 May 2011 15:02:02 +0200 MIME-Version: 1.0 Sent to CCL by: Andreas Klamt [klamt!^!cosmologic.de] Hi John, hi Pierre, I need to contradict to some degree. As I mentioned several times, the vibrational, rotational and translational contributions in solutions are very hard to quantify due to the coupling of molecules. Hence I do not know where a value as 30% of gas-phase value in water comes from. I guess, nobody argues that the vibrational, rotational and translational contributions are small. The argument only is that they are empirically parameterized into all currently available solvation models, because these are parameterized against exp. data which include these effects. The situation is discussed in J Phys Chem A. 2010 Dec 30;114(51):13442-4. Comment on the correct use of continuum solvent models. Ho J, Klamt A, Coote ML. Regards Andreas Am 24.05.2011 13:42, schrieb Pierre Archirel pierre.archirel- -u-psud.fr: > Sent to CCL by: "Pierre Archirel" [pierre.archirel^-^u-psud.fr] > Dear John, > If you are calculating a difference the question is much simpler because entropy effects cancel out. Anyway: > 1- Considering that translation and rotation are simply quenched in solution is rather true for energies but completely wrong for entropies. In the vacuum the Sackur-Tetrode and rigid rotor formulas show that the TS contributions amount to (roughly) 0.4 eV (translation) and 0.4 eV (rotation) for a middle size molecule (these values only depend on the log of the mass), 0.8 eV in total therefore. In a condensed phase these contributions may take all the values between zero and the vacuum values. Zero is a plausible value in solids, I consider that in water 30% of the vacuum values is plausible, this amounts to .25 eV, this is not small! > 2- But you are looking for a difference! Your two isomers have the same mass and presumably inertial moments, close to each other. So the (large) entropy contributions from translation and rotation nicely cancel out and your formula DeltaG (in solution) = (G(B,gas_vib) + DeltaG(B,solv)) - (G(A,gas_vib) + DeltaG(A,solv)) eventually quite correct. > 3- as for vibrations their effect can be easily evaluated through harmonic analysis in the vacuum. > 4- I emphasize that translation and rotation entropies are not small in usual solvents. They are considered zero by people who do not know how to calculate them. > > Pierre Archirel > LCP, Universite Paris-Sud > France > pierre.archirel|u-psud.fr > > > -- PD. Dr. Andreas Klamt CEO / Geschäftsführer COSMOlogic GmbH& Co. KG Burscheider Strasse 515 D-51381 Leverkusen, Germany phone +49-2171-731681 fax +49-2171-731689 e-mail klamt*o*cosmologic.de web www.cosmologic.de HRA 20653 Amtsgericht Koeln, GF: Dr. Andreas Klamt Komplementaer: COSMOlogic Verwaltungs GmbH HRB 49501 Amtsgericht Koeln, GF: Dr. Andreas Klamt From owner-chemistry@ccl.net Tue May 24 15:23:01 2011 From: "Pierre Archirel pierre.archirel{}u-psud.fr" To: CCL Subject: CCL: Solvation free energies Message-Id: <-44739-110524105004-18243-o/3LJNpzR+8/BpbyMYDK7A|server.ccl.net> X-Original-From: "Pierre Archirel" Date: Tue, 24 May 2011 10:50:02 -0400 Sent to CCL by: "Pierre Archirel" [pierre.archirel^^u-psud.fr] Dear John, dear Andreas, I recently stated that translation-rotation entropies in water amounts to roughly 30% of the values in the vacuum. I now explain this statement, though I have not the numbers under the hand: 1- Str(vac, 1atm) is the translation-rotation entropy of H2O in the gas phase (Sackur-Tetrode + rigid rotor), you may remove -RLn 24.5 for having water gazeous and 1M. 2- the experimental value of the standard molar entropy of liquid water is known: S^0 (for water 55.5M) 3- if you add +RLn (55.5) to S^0 (for having water in solution 1M in itself) and remove the vibration entropy (calculated with harmonic analysis), then you get a number roughly 30% of the 1M gas phase value. I conclude that passing from 1M gas phase to 1M solution (in itself), water molecules undergo a 70% decrease of their translation-rotation entropy. I acknowledge that I here assume that vibration can be separated from other motions, anyway the same reasoning is possible for the whole translation-rotation-vibration entropy. Pierre Archirel LCP universite Paris-Sud France pierre.archirel|u-psud.fr From owner-chemistry@ccl.net Tue May 24 17:16:00 2011 From: "Nikita Basant nikita.technocrat06 . gmail.com" To: CCL Subject: CCL: vROCS help!! Message-Id: <-44740-110524163147-24593-O9+p9lcyjwp4GuSk7vJ/XA . server.ccl.net> X-Original-From: "Nikita Basant" Date: Tue, 24 May 2011 16:31:43 -0400 Sent to CCL by: "Nikita Basant" [nikita.technocrat06]-[gmail.com] Hey I am working with VRocs as a tool for VS for my dataset, but somehow the software dosent perform the VS very efficiently, by picking up the actives and decoys both in the validation run. Could some one tell me how to improve the efficiency of the software. I have also worked with the parameter file, by changing the tanimoto cutoff and score, could some one tell me what would be the best range for both the cutoffs to make sure that the software dosent pick up the decoys in the validation run. Any advice would be appreciated!! Thanks in advance From owner-chemistry@ccl.net Tue May 24 18:44:00 2011 From: "Paul Hawkins phawkins*eyesopen.com" To: CCL Subject: CCL: vROCS help!! Message-Id: <-44741-110524184245-18659-bP80qyjeyaNQQXdqqsnzsg]^[server.ccl.net> X-Original-From: Paul Hawkins Content-Language: en-US Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset="us-ascii" Date: Tue, 24 May 2011 15:42:38 -0700 MIME-Version: 1.0 Sent to CCL by: Paul Hawkins [phawkins=eyesopen.com] Nikita, Since you have already written directly to me and to support here at OpenEye, and we have replied, you should already have a response to these questions. However since the reply may be of general interest to the list I copy below the answer I have already sent you. Some general comments about virtual screening: It is a probabilistic ranking method, the goal being to score the actives higher than the decoys/inactives in a ranked list. No virtual screening tool will score all the possible actives higher than all the possible inactives, so you should expect that ROCS, and any other tool, will give a good score/high rank to inactives/decoys from time to time. Bear in mind that any similarity method, be it 3D like ROCS or 2D like fingerprints, will always make mistakes - small changes in structure can result in large changes in activity, and these cases are unlikely to be detected by any similarity method. Using score cutoffs can be tricky in any VS experiment as the main goal is ranking the molecules. If you have confidence that you have a representative sampling of all the actives that could possibly be interesting to you and good decoys then you can experiment with score cutoffs, but this does require careful thought. The danger is that you will miss interesting molecules that might receive a reasonable rank but a relatively poor score. A good idea is to plot a rank-score plot for a database and see where the actives come. Some questions about your vROCS use: How did you choose the decoys and the actives? The nature of the actives and decoys does have a strong influence on the reliability of a validation. How many of each are you using? How did you generate the conformations for the actives and decoys that you are using? In general the way you are using vROCS seems very sensible. Using a single molecule as a query, in a conformation derived from experimental data, as in your case, or even one calculated from OMEGA, often gives good results with ROCS, but for any individual case ROCS' performance is, obviously, impossible to predict. Paul Hawkins. Senior Applications Scientist OpenEye Scientific Software -----Original Message----- > From: owner-chemistry+phawkins==eyesopen.com^_^ccl.net [mailto:owner-chemistry+phawkins==eyesopen.com^_^ccl.net] On Behalf Of Nikita Basant nikita.technocrat06 . gmail.com Sent: Tuesday, May 24, 2011 1:32 PM To: Paul Hawkins Subject: CCL: vROCS help!! Sent to CCL by: "Nikita Basant" [nikita.technocrat06]-[gmail.com] Hey I am working with VRocs as a tool for VS for my dataset, but somehow the software dosent perform the VS very efficiently, by picking up the actives and decoys both in the validation run. Could some one tell me how to improve the efficiency of the software. I have also worked with the parameter file, by changing the tanimoto cutoff and score, could some one tell me what would be the best range for both the cutoffs to make sure that the software dosent pick up the decoys in the validation run. Any advice would be appreciated!! Thanks in advancehttp://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txt