From owner-chemistry@ccl.net Wed May 31 07:08:00 2006 From: "Marcin Makowski makowskm=-=chemia.uj.edu.pl" To: CCL Subject: CCL: negative bsse content Message-Id: <-31869-060531070632-22347-Y3QEVQEVoowf8wwUTYbDkQ.:.server.ccl.net> X-Original-From: "Marcin Makowski" Date: Wed, 31 May 2006 07:06:28 -0400 Sent to CCL by: "Marcin Makowski" [makowskm-x-chemia.uj.edu.pl] Very general thing is that actual LMP2 implementation is an approximation to exact MP2 relying (by cut-offs introduced) on localization of orbitals. In this method and in hte limit of perfect localization BSSE should just disappear. For this reason BSSE here has good chance to be quite small. I wouldn't be suprised if subtle changes in the shape of localized orbitals and effective partition of space between separate computations could actually transform some small positive net effect into small negative. The result is not physical, but I wouldn't worry too much about it if the values are under 1 kcal. If you want to get some more confidence, try decreasing cut-off tresholds of the LMP2 method and see if you obtain smooth behaviour. Yours, Marcin > Sent to CCL by: Jozsef Csontos [jozsefcsontos[a]creighton.edu] > Dear List Members, > > I was calculating intermolecular interaction energies using the > counterpoise method to correct the results for bsse (3 dimer centered > basis set calcs). The bsse contents were also calculated (2 more monomer > centered basis set calcs). If I'm not mistaken I can get the bsse > content if I subtract the non-bsse corrected energy from the bsse > corrected one. This imply that the bsse content should be positive. > > I am using quite large basis set considering the investigated systems > (~50 atoms, cc-pVTZ, aug-cc-pVTZ) and the BSSE content is usually small, > less than 1kcal/mol; the interaction energies are in the 2-8 kcal/mol > range. The level of theory are lmp2 and dft. > > However, in some cases (mainly lmp2 calcs) I got small negative (~ > -0.3,-0.5 kcal/mol) bsse contents. > > Do you have any idea, how the bsse content can come to be negative? > > Thanks, > Jozsef > > > -- > Jozsef Csontos, Ph.D. > > Department of Biomedical Sciences > Creighton University, > Omaha, NE From owner-chemistry@ccl.net Wed May 31 07:49:01 2006 From: "Tanja van Mourik tanja.vanmourik%x%st-andrews.ac.uk" To: CCL Subject: CCL: negative bsse content Message-Id: <-31870-060531064727-21144-Y+dZXlzaw/ecsJLLHJWbQQ-x-server.ccl.net> X-Original-From: Tanja van Mourik Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Wed, 31 May 2006 11:11:22 +0100 MIME-Version: 1.0 Sent to CCL by: Tanja van Mourik [tanja.vanmourik^-^st-andrews.ac.uk] Dear Joszef, > > I was calculating intermolecular interaction energies using the > counterpoise method to correct the results for bsse (3 dimer centered > basis set calcs). The bsse contents were also calculated (2 more monomer > centered basis set calcs). If I'm not mistaken I can get the bsse > content if I subtract the non-bsse corrected energy from the bsse > corrected one. This imply that the bsse content should be positive. > > I am using quite large basis set considering the investigated systems > (~50 atoms, cc-pVTZ, aug-cc-pVTZ) and the BSSE content is usually small, > less than 1kcal/mol; the interaction energies are in the 2-8 kcal/mol > range. The level of theory are lmp2 and dft. > > However, in some cases (mainly lmp2 calcs) I got small negative (~ > -0.3,-0.5 kcal/mol) bsse contents. > > Do you have any idea, how the bsse content can come to be negative? BSSE is a negative quantity: it is calculated as the difference between the sum of the monomer energies calculated in the dimer basis set and the sum of the monomer energies calculated in the monomer basis set, all of these at the geometries they adopt in the complex. It can also be calculated by subtracting the CP-corrected interaction energy from the uncorrected interaction energy (i.e., other way around compared to what you say above), but -only- when the deformation energies are included in the CP-corrected interaction energy. Perhaps some formulas can help (with latex-style sub- and superscripts): BSSE = E_{A}^{AB}(AB) + E_{B}^{AB}(AB) - E_{A}^{A}(AB) - E_{B}^{B}(AB) where the subscripts, _{A} and _{B}, denote the molecular systems; the superscipts, ^{A}, ^{B} and ^{AB}, denote the basis set (monomer or dimer centred basis sets), and the (AB) in round brackets denotes that all these are calculated at the optimised geometry of the dimer AB. Using the same notation: DeltaE(CP) = E_{AB}^{AB}(AB) - E_{A}^{AB}(AB) - E_{B}^{AB}(AB) + E_{A}^{A}(AB) + E_{B}^{B}(AB) - E_{A}^{A}(A) - E_{B}^{B}(B) DeltaE(noCP) = E_{AB}^{AB}(AB) - E_{A}^{A}(A) - E_{B}^{B}(B) where E_{A}^{A}(A) is the energy of A at the equilibrium geometry of A, (A), calculated in the monomer basis set. The deformation energy of monomer A is: Edef_{A} = E_{A}^{A}(AB) - E_{A}^{A}(A) Comparing DeltaE(CP) and Delta(noCP) you can see that the difference between these two equation is just the definition of the BSSE: DeltaE(CP) = DeltaE(noCP) - BSSE It sounds to me that you are calculating the BSSE as: BSSE = E_{A}^{AB}(AB) + E_{B}^{AB}(AB) - E_{A}^{A}(A) - E_{B}^{B}(B) (or with opposite sign), in which case you may get positive as well as negative results (depending on how large the monomer deformation energies are). The difference between this wrong definition of BSSE and the correct BSSE is the sum of the monomer deformation energies. For the formulas, see also: Adv. Quant. Chem. 31, pp 105-135 (1999) or less elaborately in some of my other papers, for example: Phys. Chem. Chem. Phys. 5, pp 4519-4526 (2003) Hope this helps, Tanja -- ================================================================= Tanja van Mourik Royal Society University Research Fellow School of Chemistry, University of St. Andrews North Haugh, St. Andrews Fife KY16 9ST, Scotland (UK) email: tanja.vanmourik-#-st-andrews.ac.uk web: http://chemistry.st-and.ac.uk/staffmember.php?id=tvm ================================================================= From owner-chemistry@ccl.net Wed May 31 11:00:00 2006 From: "Jan Labanowski janl(-)speakeasy.net" To: CCL Subject: CCL: CCL Moderator reminders and appeals Message-Id: <-31871-060531105448-31908-qBz0ys9oMKOkHPBYessrUg-x-server.ccl.net> X-Original-From: "Jan Labanowski" Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="iso-8859-1" Date: Wed, 31 May 2006 14:54:39 +0000 MIME-Version: 1.0 Sent to CCL by: "Jan Labanowski" [janl*speakeasy.net] Dear CCL Members, The vacations time is approaching, so few reminders: 0) When you send a message to CCL remember that it will be posted ONLY after you confirm it (you will get a confirmation request by e-mail -- if it does not show up, check your spambox!). 1) Advertise your job openings before students/post-docs are gone! If you are a group member, tell your boss about the Job Service at: http://www.ccl.net/jobs It is a very popular/successful service and will bring you candidates! Please spread the word about it among those who hire and those who look for a job and you will be doing them a favor. 2) Review and advertise conferences! It is easy and free (but your support is very WELCOME!). Go to: http://server.ccl.net/chemistry/announcements/conferences Submit interesting conferences/workshop/symposium there even if you are not associated with the conference. 3) If you are leaving for prolonged vacations and your mailbox quota is small, please consider unsubscribing from CCL at: http://server.ccl.net/chemistry/sub_unsub.shtml You can always resubscribe when you return, and you can browse the messages that you missed on the Web at: http://server.ccl.net/chemistry/resources/messages 4) Please upload materials to CCL Archives http://server.ccl.net/chemistry/aboutccl/contributing You will get deserved recognition and your materials will be noticed by Computational Chemistry community. 5) Please use CCL Services and support CCL. CCL is run from a private home and is supported ONLY by YOU. Be one of CCL Supporters! http://server.ccl.net/chemistry/aboutccl/supporting Consider "Adopting a CCL Page" by your company or your group. It will increase your web traffic and provide you with the beneficial publicity as a CCL Supporting organization. For all paid services you will get a regular Invoice, and you can easily pay with a Credit Card on line or by other methods, like company check or wire transfer. 6) Consider ordering your copy of CCL Archives on DVD. It will allow you to grep/find/awk/perl/whatever the files on the DVD on your local computer rather than do it via Internet. 7) CCL.NET is an old and recognized domain. It is indexed by all major search engines. Tell me how CCL can help you in promoting your web site, software, consulting, etc. You will capitalize on the CCL established status on the Internet and support CCL at the same time. I will help you if you help CCL. Note... There are so many companies that service Computational Chemistry, Cheminformatics, Bioinformatics, etc., and so on... Ask them to support CCL! You may have learned about them by looking at list of CCL Links: http://server.ccl.net/chemistry/links/index.shtml yet, they do not support CCL beside the http://nanoandgiga.com and the http://sunsetmolecular.com/ that "Adopted the CCL Page" and the ones listed on Supporters Page at: http://server.ccl.net/cgi-bin/ccl/supporting_members Thank you for your attention... Your CCL Moderator Jan Labanowski, Ph.D. jkl : ccl.net From owner-chemistry@ccl.net Wed May 31 11:34:01 2006 From: "manuel alonso tarajano%x%bioinfo.cu" To: CCL Subject: CCL: problems with charged residues at MDyn Message-Id: <-31872-060531103810-27707-g9sKEnK9uqFRADQRa4d2bg++server.ccl.net> X-Original-From: manuel alonso Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Wed, 31 May 2006 09:42:22 -0400 MIME-Version: 1.0 Sent to CCL by: manuel alonso [tarajano%%bioinfo.cu] hello list: (first of all .. im very new at charmm .. so i hope you will apologize any tribial question) i know you have had a lot of disscusions here about protonation states of a.acids residues ..but till now I havent found the one that could help me Im running (trying to) a MDyn to a small peptide with titratable groups and I think CHARMM is complaining about its protonations state and cos of this my MDyns runs are dying. Here I post part of the output I think is more related to the problem : . . . . (here are my titratable residues) . ***** Message from SEQRDR ***** THE SYSTEM CONTAINS 9 TITRATABLE GROUPS THE USER MUST PREDETERMINE THE PROTONATION STATE THROUGH THE SEQUENCE AND RTF HIS - 0 ASP - 3 GLU - 2 LYS - 4 TYR - 0 . . . . (here are the warnings) . . . TITLE> * ** WARNING ** For atom in coordinate file, the residue type does not match that (RESN) in the PSF: 2 PSF= THR INPUT= ASN ** WARNING ** For atom in coordinate file, the residue type does not match that (RESN) in the PSF: 2 PSF= THR INPUT= ASN ** WARNING ** For atom in coordinate file, the residue type does not match that (RESN) in the PSF: 2 PSF= THR INPUT= ASN ** WARNING ** For atom in coordinate file, the residue type does not match that (RESN) in the PSF: 2 PSF= THR INPUT= ASN ** WARNING ** For atom in coordinate file, the residue type does not match that (RESN) in the PSF: 2 PSF= THR INPUT= ASN ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 6 IRES= 2 RESID=3 RES=ASN ATOM=CG ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 7 IRES= 2 RESID=3 RES=ASN ATOM=OD1 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 8 IRES= 2 RESID=3 RES=ASN ATOM=ND2 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 14 IRES= 3 RESID=4 RES=THR ATOM=OG1 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 15 IRES= 3 RESID=4 RES=THR ATOM=CG2 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 22 IRES= 4 RESID=5 RES=ASP ATOM=OD1 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 23 IRES= 4 RESID=5 RES=ASP ATOM=OD2 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 40 IRES= 6 RESID=7 RES=GLN ATOM=OE1 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 41 IRES= 6 RESID=7 RES=GLN ATOM=NE2 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 47 IRES= 7 RESID=8 RES=LYS ATOM=CG ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 48 IRES= 7 RESID=8 RES=LYS ATOM=CD ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 49 IRES= 7 RESID=8 RES=LYS ATOM=CE ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 50 IRES= 7 RESID=8 RES=LYS ATOM=NZ ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 56 IRES= 8 RESID=9 RES=VAL ATOM=CG1 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 57 IRES= 8 RESID=9 RES=VAL ATOM=CG2 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 63 IRES= 9 RESID=10 RES=SER ATOM=OG ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 69 IRES= 10 RESID=11 RES=GLU ATOM=CG ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 71 IRES= 10 RESID=11 RES=GLU ATOM=OE1 ** WARNING ** For atom in coordinate file, the corresponding residue in the PSF lacks that atom: INDEX= 72 IRES= 10 RESID=11 RES=GLU ATOM=OE2 ** WARNING ** For atom in coordinate file, could not find residue in PSF, and is thus ignored: SEGID= RESID=32 RESNAME= MET TYPE= N *** LEVEL 1 WARNING *** BOMLEV IS 0 ** WARNING ** For atom in coordinate file, could not find residue in PSF, and is thus ignored: SEGID= RESID=32 RESNAME= MET TYPE= CA *** LEVEL 1 WARNING *** BOMLEV IS 0 ** WARNING ** For atom in coordinate file, could not find residue in PSF, and is thus ignored: SEGID= RESID=32 RESNAME= MET TYPE= C *** LEVEL 1 WARNING *** BOMLEV IS 0 ** WARNING ** For atom in coordinate file, could not find residue in PSF, and is thus ignored: SEGID= RESID=32 RESNAME= MET TYPE= O *** LEVEL 1 WARNING *** BOMLEV IS 0 ** WARNING ** After reading, there are no coordinates for selected atom: 1 1 ASN HT1 ** WARNING ** After reading, there are no coordinates for selected atom: 2 1 ASN HT2 ** WARNING ** After reading, there are no coordinates for selected atom: 3 1 ASN N ** WARNING ** After reading, there are no coordinates for selected atom: 4 1 ASN HT3 ** WARNING ** After reading, there are no coordinates for selected atom: 5 1 ASN CA ** WARNING ** After reading, there are no coordinates for selected atom: 6 1 ASN CB ** WARNING ** After reading, there are no coordinates for selected atom: 7 1 ASN CG ** WARNING ** After reading, there are no coordinates for selected atom: 8 1 ASN OD1 ** WARNING ** After reading, there are no coordinates for selected atom: 9 1 ASN ND2 ** WARNING ** After reading, there are no coordinates for selected atom: 10 1 ASN HD21 ** A total of 133 selected atoms have no coordinates ** A total of 64 warnings were encountered during coordinate reading ** ** MESSAGE ** 8 atoms in coordinate file were outside the specified sequence range. *** LEVEL 2 WARNING *** BOMLEV IS 0 ** WARNING ** 204 atoms in coordinates file had a sequence mismatch. *** LEVEL 0 WARNING *** BOMLEV IS 0 BOMLEV HAS BEEN SATISFIED. TERMINATING. Im currently using CHARMM 30b1 I guess im missing to supply CHARMM with some PSF file/data related to titratable groups, but I dont know which file might this be or if I should provide CHARMM with some info to handle titratable groups at a given pH. I think I remeber that for example histidine (HIS) has two different 3-letters-code for its two protonations states , maybe i must provide charmm with the rights 3-letters-codes for my titrable groups ? if so.. where could i find those codes ? thanks to all in advance and please let me know if you need some more data in order to help me tarajano (the initial part of my input file is:) ********************************************************** ********************************************************** open unit 1 card read name toph19_eef1.inp read RTF card unit 1 close unit 1 open unit 1 card read name param19_eef1.inp read PARA card unit 1 close unit 1 open unit 1 card read name "xxxxxx.pdb" read sequ pdb unit 1 generate prot setu ! pdb open unit 1 card read name "xxxxxx.pdb" read coor pdb unit 1 close unit 1 ! hydrogens hbuild sele all end ! build ic fill ic para ic build open unit 1 card write name "xxxxxx.pdb" write coor pdb unit 1 close unit 1 ********************************************************** ********************************************************** From owner-chemistry@ccl.net Wed May 31 12:56:00 2006 From: "Rick Venable rvenable.:.pollux.cber.nih.gov" To: CCL Subject: CCL: problems with charged residues at MDyn Message-Id: <-31873-060531125347-30107-cyRsZ4MLWnOqsFuH6qNW9w~!~server.ccl.net> X-Original-From: Rick Venable Content-Type: TEXT/PLAIN; charset=US-ASCII Date: Wed, 31 May 2006 12:49:45 -0400 MIME-Version: 1.0 Sent to CCL by: Rick Venable [rvenable/a\pollux.cber.nih.gov] The CHARMM forums at URL www.charmm.org may be a better choice for this sort of question, and the web site does support the use of attachments (CCL does not, for understandanle email security reasons). The 'titratable group' message is warning only, a reminder to carefully review choices made about things such as the location and number of protons on the HIS ring. There are likely other errors. On Wed, 31 May 2006, manuel alonso tarajano%x%bioinfo.cu wrote: > (first of all .. im very new at charmm .. so i hope you will apologize > any tribial question) i know you have had a lot of disscusions here > about protonation states of a.acids residues ..but till now I havent > found the one that could help me Im running (trying to) a MDyn to a > small peptide with titratable groups and I think CHARMM is complaining > about its protonations state and cos of this my MDyns runs are dying. > Here I post part of the output I think is more related to the problem ------------------------------------- Rick Venable 29/500 Membrane Biophysics Section NIH/NHLBI Lab. of Comp. Biology Bethesda, MD 20892-8014 U.S.A. (301) 496-1905 Rick_Venable AT nih*gov ALT email: rvenable AT speakeasy*org ------------------------------------- From owner-chemistry@ccl.net Wed May 31 13:32:01 2006 From: "Eric Hu list.eric^gmail.com" To: CCL Subject: CCL: freeze internal coordinates in G03 Message-Id: <-31874-060531132933-22940-rFdwrv37CCmUMecNbYd9AQ^_^server.ccl.net> X-Original-From: "Eric Hu" Content-Type: multipart/alternative; boundary="----=_Part_15212_9871662.1149096565091" Date: Wed, 31 May 2006 10:29:25 -0700 MIME-Version: 1.0 Sent to CCL by: "Eric Hu" [list.eric]|[gmail.com] ------=_Part_15212_9871662.1149096565091 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Content-Disposition: inline Hi, I want to freeze two monomers in a dimer structure during a geometry optimization. ModRedundant provides several ways to freeze bond length, angle and dihedral angles. Since I want to freeze any internal geometry optimization within the monomer and am only interested in their relative orientation in the dimer, I wonder if there are more concise ways to deal with this besides using B ** F, A***F and D***F for every atom. Thanks. Eric ------=_Part_15212_9871662.1149096565091 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline
Hi, I want to freeze two monomers in a dimer structure during a geometry optimization. ModRedundant provides several ways to freeze bond length, angle and dihedral angles. Since I want to freeze any internal geometry optimization within the monomer and am only interested in their relative orientation in the dimer, I wonder if there are more concise ways to deal with this besides using B ** F, A***F and D***F for every atom. Thanks.

Eric ------=_Part_15212_9871662.1149096565091-- From owner-chemistry@ccl.net Wed May 31 14:58:00 2006 From: "Jozsef Csontos jozsefcsontos|a|creighton.edu" To: CCL Subject: CCL: negative bsse content Message-Id: <-31875-060531145149-23464-wVyHGiFQEfUrpTWJ8TpocA{:}server.ccl.net> X-Original-From: Jozsef Csontos Content-Transfer-Encoding: 7bit Content-Type: text/plain Date: Wed, 31 May 2006 13:51:39 -0500 Mime-Version: 1.0 Sent to CCL by: Jozsef Csontos [jozsefcsontos%%creighton.edu] Hi Tanja, > Dear Joszef, > > BSSE is a negative quantity: it is calculated as the difference between > the sum of the monomer energies calculated in the dimer basis set and > the sum of the monomer energies calculated in the monomer basis set, all > of these at the geometries they adopt in the complex. > Perhaps some formulas can help > (with latex-style sub- and superscripts): > > BSSE = E_{A}^{AB}(AB) + E_{B}^{AB}(AB) - E_{A}^{A}(AB) - E_{B}^{B}(AB) you are right, equations can help, sorry for neglecting them. I calculated the interaction energy according to the original Boys-Bernardi procedure: DeltaE(CP)=E_{AB}^{AB}(AB) - E_{A}^{AB}(AB) - E_{B}^{AB}(AB) in this case the non corrected interaction energy is DeltaE(noCP) = E_{AB}^{AB}(AB) - E_{A}^{A}(A) - E_{B}^{B}(B) The difference between the two is the bsse content, which is the same as you stated above. Some words about the sign, I think the sign of the bsse is the matter of definition: (+) JCC, 25, 1771, 2004 (-) JCC, 22, 196, 2001 I used the DeltaE(CP)-DeltaE(noCP) equation which implies that the bsse is positive. This gave me results with opposite sign comparing to yours: > DeltaE(CP) = DeltaE(noCP) - BSSE > As you might noticed, I didn't correct for deformation (fragment relaxation), but it vanishes anyway when you subtract the interaction energies. So, apart from the signs, I should always get results on the same side of zero, but I didn't. Best wishes, Jozsef PS.: I tend to accept Marcin's opinion. >I wouldn't be suprised if subtle changes in the shape of localized >orbitals and effective partition of space between separate computations >could actually transform some small positive net effect into small >negative. -- Jozsef Csontos, Ph.D. Department of Biomedical Sciences Creighton University, Omaha, NE From owner-chemistry@ccl.net Wed May 31 17:25:00 2006 From: "Serguei Patchkovskii ps.:.ned.sims.nrc.ca" To: CCL Subject: CCL: negative bsse content Message-Id: <-31876-060531172244-31239-H4GfxyatwKBbp35/necG+g . server.ccl.net> X-Original-From: Serguei Patchkovskii Content-Type: TEXT/PLAIN; charset=US-ASCII Date: Wed, 31 May 2006 16:41:53 -0400 (EDT) MIME-Version: 1.0 Sent to CCL by: Serguei Patchkovskii [ps,ned.sims.nrc.ca] On Wed, 31 May 2006, Jozsef Csontos jozsefcsontos|a|creighton.edu wrote: > BSSE = E_{A}^{AB}(AB) + E_{B}^{AB}(AB) - E_{A}^{A}(AB) - E_{B}^{B}(AB) [...] > So, apart from the signs, I should always get results on the same side > of zero, but I didn't. Given the above definition, you can expect BSSE to be non-positive as long as the method used to calculate the energies is variational. The argument goes as follows: A bigger variational space (A in the combined A+B basis set) can only lead to lower energies compared a smaller variational space (A in the A basis set alone). Therefore, the difference E_{A}^{AB}(AB)-E_{A}^{A}(AB) must be negative (or zero). Same holds for the energy of B - so the sum of the two quantities must be non-positive as well. This argument breaks down if: a) the method you are using is not variational or b) the basis set alone does not represent the entire variational space of the method or c) there is numerical noise in the calculations, which exceeds the magnitude of the BSSE Because MP2 energy is not variational, there is no reason to expect a definite sign of BSSE for MP2 (either canonical or localized). In the case of DFT, there is both a "hidden" variational space (numerical integration grid) and a possibility of having significant numerical noise. Both effects should decrease with the use of better grids. Serguei