From STINCH@IMISIAM.bitnet Mon Mar 15 11:02:59 1993 Message-Id: <199303151555.AA22445@oscsunb.ccl.net> Date: Mon, 15 Mar 93 16:48:02 ITA From: Davide Proserpio Subject: Ribbons---> molscript ?? To: CHEMISTRY@ccl.net Hi, Please anyone knows of any software to draw protein molecules in Ribbons? I saw something called Molscript (probably), anyone has some information about it or about something else? Thank you for the help. Davide **************************************************************** Davide M. Proserpio UNIVERSITA' DI MILANO ISTITUTO DI CHIMICA STRUTTURISTICA INORGANICA VIA VENEZIAN 21, 20133 MILANO - ITALY PHONE : 39-2-70635120 FAX : 39-2-70635288 E-MAIL: STINCH@IMISIAM.MI.CNR.IT **************************************************************** From wrinn@iris104.biosym.com Mon Mar 15 00:08:19 1993 Date: Mon, 15 Mar 93 08:08:19 -0800 From: wrinn@biosym.com (Michael Wrinn) Message-Id: <9303151608.AA01039@iris104.biosym.com> To: cundarit@memstvx1.memst.edu Subject: more on LDF followup Tom, Thanks for posting the summary of responses to your LDF questions; let me toss in another two cents: > First question: Are LDF methods size-consistent? One of the reality checks we do for program implementations is to calculate, say, an N2 molecule, then 2 N2 molecules at large (1000 au) distance apart - these had better be exactly a factor of 2 apart in energy. (Note, however, that grid noise in numeric methods (DMol included) implies a tuneable threshold of "exactness".) > Third question: I have just calcd. a binding energy of -30 kcal/mol > from E(Cr(CO)5(H2)) - E(Cr(CO)5) -E(H2) using the equivalent of a double-zeta- > valence-plus-polarization basis set, and a FINE mesh size. Should I believe > this or is it worth the expense to go to a larger basis set, finer mesh, > unfrozen core? The overwhelming improvement, as stated by several contributors, comes from using gradient-corrected functionals. That said, a somewhat larger basis does improve binding energy results. We've found that using 2 polarization functions (DNPP, in DMol jargon) comes close to saturating the basis (coincidentally giving results within a few tenths of kcal of Axel Becke's basis-set-free numbers (JCP 96, 2155) which were taken from a hartree-fock density); f functions move this slightly further, g and higher functions have almost no influence. Although the DMol basis set nomenclature resembles "double-zeta plus polarization", please be aware that the basis sets used are numeric atomic functions (exact, at the same level of theory as the molecular calculations), and thus are not equivalent to STO. regards, Michael Wrinn From sauer@organik.uni-erlangen.de Mon Mar 15 13:02:55 1993 From: Wolfgang Sauer Message-Id: <9303151713.AA14364@derioc1.organik.uni-erlangen.de> Subject: AMINO ACIDS by DFT (followup) To: CHEMISTRY@ccl.net (CompChem MailExploder) Date: Mon, 15 Mar 93 18:13:03 MET Original message: > From: siroiss@cerca.umontreal.ca (Suzane Sirois) > Subject: AMINO ACIDS > Date: Fri, 12 Mar 1993 17:46:32 -0500 (EST) > > I would like to have references for the bond length > the infrared spectra and the dipole moment of amino acids > that i can compared with DFT calculation > a more general question to the DFT gurus out there: how good are DFT calculations on H bonded systems like amino acids or small (or even larger ?) peptides ? References, general remarks, practical hints, ... all welcome. Thanks, Wolfgang. PS: I'll summarize for the list, if answers are mailed directly to me. -- +======================================+=====================================+ | Wolfgang Sauer | | | Institut fuer Organische Chemie I | "You have to be three standard | | Henkestr. 42, D-8520 Erlangen, FRG | deviations away from a normal | | | personality to like UNIX." | | sauer@organik.uni-erlangen.de | | | Tel.: 49/0 - 9131 - 85 - 2952 | Morris Jones (C&T) | | Fax: - 9132 | | +======================================+=====================================+ From UDIM018@FRORS31.bitnet Mon Mar 15 13:39:45 1993 Message-Id: <199303151754.AA24425@oscsunb.ccl.net> Date: 15 Mar 93 17:39:45 EDT From: Subject: partial charges for fictive atoms in G92 To: chemistry@ccl.net E. M. EVLETH Dynamique des Interactions Moleculaires Universite Pierre et Marie Curie 4 Place Jussieu, Tour 22, Paris 75005 33-1-44-27-42-08 (work), 33 = France; 1 = Paris 33-1-45-48-67-20 (home) FAX 33-1-44-27-42-17 e-mail UDIM018 at FRORS31.BITNET Thanks for Mike Frisch for his clarifications. Having G92 version B, I am wondering about what versions beyond that offer as corrections? Is it possible to charge in fictive atoms have non-integral charges? in Gaussian 92 (the manual indicates probably not). Why is this needed? When modeling inorganic crystals (having for instance hexacorrdinated Al) terminating the crystal can not be accomplished using integral charges on fictive atoms. The alternative choice is to use a periodic HF technique (which we are doing) but this method has its inconveniences if one is treating non-periodic systems. Likewise one can model in a zeolite those portions of the system not taken into the quantum modeling using fixed partial charges. R. Poirier's MUNGAUSS program which has been programmed to do this. The sparkles option in AMPAC and MOPAC use integral charges (although introducing fractional charges is being contemplated). But generally this possible useful (possibly dangerous) option is not available. From HSUS@chemv2.mps.ohio-state.edu Mon Mar 15 07:55:00 1993 Date: Mon, 15 Mar 1993 12:55 EST From: HSUS@chemv2.mps.ohio-state.edu Subject: Problem with vibrational frequency calculation using GAUSSIAN92 To: CHEMISTRY@ccl.net Message-Id: Dear Everybody, I would appreciate very much for your reply to solve my problem. Currently I am studying the [B10H14I]- ion (which has C2v symmetry) using the GAUSSIAN92. Especially I need to calculate its vibrational frequencies. After optimizing its geometry using LANL1DZ basis set with (and without) tight (or verytight) option, G92 seems to have trouble to calculate its frequencies. Because after few cycles, it lost the right orbital symmetries and cycles are repeated (infinitively?) In order to find out what the problem is, 1st, I calculated I2 molecule's frequency using the LANL1DZ. G92 took 10 cycles, without losing its orbital symmetries, and finished its frequency calculation. Then I used both D95V and LANL1DZ basis sets for molecule B10H14 (which also has C2v symmetry) and calculate its frequencies. It turns out that the D95V worked for frequency calculation, but not LANL1DZ. When G92 used the D95V basis set, it computed the 2nd derivatives of energy analytically. Therefore, there is no iteration cycles and it only took two hours of CPU to finish frequency calculation. However, when it used the LANL1DZ, G92 will calculate the 2nd derivatives of energy numerically. After 2 cycles, many OCCUPIED and VIRTUAL ORBITAL SYMMETRIES became either (?A) or (?B) or (?C). So far, the calculation is at the 98 cycles and is still repeating (for each cycle, the unknown orbital symmetries are different). Each cycles takes about 68 minutes of CPU (total it has taken more than 5 days). If anyone has this experience? If this is the bug of G92? If it is not a bug of G92, then how many cycles will it normally take for a molecule in this size? If it is a bug of G92, then how one can work around to get results? Thank you very much in advance to anyone who can help me. Sincerely yours, Ruth Hsu From ZSYAMP01@EBCESCA1.bitnet Mon Mar 15 13:35:34 1993 Message-Id: <199303151755.AA24442@oscsunb.ccl.net> Date: Mon, 15 Mar 93 17:56:40 CET From: ZSYAMP01%EBCESCA1@OHSTVMA.ACS.OHIO-STATE.EDU Subject: help needed To: chemistry@ccl.net Dear colleagues, Here at CESCA (the Supercomputing Center of Catalonia, the EASI site in Spain) we are planning to start a new research line, involving Molecular Modelling. This was a field traditionally set by a few academic researchers, but we would like to give it a major outreach, spreading the use of supercomputers to private companies whose R+D departments would use our resources as an aid in the development of their products. I would be grateful if you could help me in the choice of the adequate products for this purpose, and maybe sharing your valuable experiences with me. So far our resources include an IBM 3090/600 (plus VF), a CRAY YMP-2E 232 and some Silicon Graphics Indigo workstations; amongst the software available we have already got GAUSSIAN, MOPAC, MOTECC, GAMESS, to cover the Quantum Chemistry field. But we have no experience with products like DISCOVER, AMBER, CHARMM, it is, Molecular Mechanics, Molecular Dynamics, Simulation techniques, and so on. The question is, 'WHICH SOFTWARE WOULD YOU RECCOMMEND ME TO INSTALL AND WHY?' (Almost everyday I ask myself questions like 'Is DISCOVER as powerful as the vendor says?', 'Are there other packages less expensive which could do the same?', 'Is UNICHEM a good choice because of the product itself or is it so just because users want its LDF program?', 'Is it worth to buy software for a mainframe when there are efficient versions for workstations?', 'What about databases?', 'And visualization?') Please, reply directly to me if you prefer not to bring this subject to public discussion in the list. I thank you in advance. Alicia Martinez System Engineer, CESCA BITNET: ZSYAMP01 at EBCESCA1 Internet: zsyamp01 at puigmal.cesca.es From ONDRECHEN@northeastern.edu Mon Mar 15 10:51:55 1993 Date: 15 Mar 1993 15:51:55 -0500 (EST) From: ONDRECHEN@northeastern.edu Subject: ATTN: ACS-NERM CONTRIBUTORS To: chemistry@ccl.net Message-Id: <01GVUCE5WNBS8Y79C3@northeastern.edu> HELP!!!!! It appears that there was some trouble with the mail utility here. If you submitted your ACS-NERM title to me by e-mail and IF YOU DID NOT GET A CONFIRMATION FROM ME, THEN PLEASE TRY AGAIN. IT APPEARS THAT SOME MESSAGES MAY HAVE BEEN LOST. (WE DID have some downtime and other anomalies in recent days, perhaps related to the storm.) If you did get a confirmation from me, then all is well. If you didn't get a confirmation, please try again, and sorry for the trouble. Mary Jo Ondrechen From st-amant@cgl.ucsf.EDU Mon Mar 15 06:01:25 1993 Date: Mon, 15 Mar 93 14:01:25 -0800 Message-Id: <9303152201.AA11229@socrates.ucsf.EDU> From: st-amant@cgl.ucsf.edu (Alain St-Amant) To: CHEMISTRY@ccl.net, sauer@organik.uni-erlangen.de Subject: Re: AMINO ACIDS by DFT (followup) >> I would like to have references for the bond length >> the infrared spectra and the dipole moment of amino acids >> that i can compared with DFT calculation >> > > a more general question to the DFT gurus out there: > how good are DFT calculations on H bonded systems like amino acids or small > (or even larger ?) peptides ? > References, general remarks, practical hints, ... all welcome. > > Thanks, > Wolfgang. here are results for the blocked alanine dipeptide that will be part of an upcoming paper dealing with the use of DFT methods in biomolecular modelling by myself, Wendy Cornell and Peter Kollman of UCSF and Tom Halgren of Merck. I think that the "message" is that local DFT methods (we used the VWN parameterization) are totally inadequate for such systems and that today's non-local functionals (we used Becke's single-parameter exchange functional and Perdew's 1986 correlation functional) provide answers that compare well with MP2. the HF and MP2 results are from Head-Gordon et. al's 1991 JACS paper. phi,psi values: HF non-local phi psi phi psi C7 equatorial -86 78 -80 65 C5 -156 160 -154 165 C7 axial 75 -54 69 -58 beta 2 -110 12 -122 27 alpha L 70 25 84 13 alpha prime -166 -41 -164 -41 and the relative conformational energies are (LDF values obtained at the non-local optimized geometries and MP2 at the HF opt. geom.): HF MP2 LDF non-local C7 equatorial 0.00 0.00 0.00 0.00 C5 0.14 1.13 2.98 1.32 C7 axial 2.53 2.19 2.79 2.07 beta 2 2.29 2.67 4.92 2.80 alpha L 4.82 4.46 5.47 3.65 alpha prime 5.69 5.83 9.33 6.53 In general, when experimental values are available (we've done a few dozen such systems spanning almost all the important functional groups) we find that non-local DFT does as well (even arguably better) as MP2 but LDF does much worse than even HF. I was also part of a paper by Sim et al. that appeard in JACS (volume 114, page 4391) that looked at some H-bonded systems. Sincerely, alain st-amant department of pharmaceutical chemistry, UCSF From tudor@wucmd.wustl.edu Mon Mar 15 14:30:23 1993 Date: Mon, 15 Mar 93 20:30:23 -0600 From: tudor@wucmd.wustl.edu (Tudor Oprea) Message-Id: <9303160230.AA00322@wucmd> To: CHEMISTRY@ccl.net Subject: Mopac >100 atoms To Mopac/Ampac programmers Those of you who have running Mopac/Ampac -s with more than 100 heavy atoms, can you please specify the ratio of Maxheavy / amount of RAM (swapspace on disk) so that I know how far could I extend my SIZES files? I can use an IBM 560 with 128 MB RAM and ~ 300 MB swapspace (and perhaps I could request something more, but not much, swapspace). Thanks to all who will bother to reply ================================================================================ = Tudor Ionel OPREA = Tel. (1-314) 935 4672 = = Research Associate = Fax. (1-314) 935 4979 = = Washington University Center for Molecular Design = Email: = = Lopata Hall, Box 1099, One Brookings Drive = tudor@wucmd.wustl.edu = = St.Louis MO 63130 = = ================================================================================ From m10!frisch@uunet.UU.NET Mon Mar 15 14:30:16 1993 Message-Id: <9303160218.AA03862@relay1.UU.NET> Date: Mon, 15 Mar 93 19:30:16 EST From: m10!frisch@uunet.UU.NET (Michael Frisch) Subject: Re: partial charges for fictive atoms in G92 To: chemistry@ccl.net E. M. EVLETH asked Thanks for Mike Frisch for his clarifications. Having G92 version B, I am wondering about what versions beyond that offer as corrections? Is it possible to charge in fictive atoms have non-integral charges? in Gaussian 92 (the manual indicates probably not). Minor revisions of Gaussian are mostly bug fixes and support for new hardware; however, we sometimes add features if the change is straightforward. In the case of scaling the frequencies, it was obvious from the discussion here that this was a useful addition, and it turned out to be quite easy. Gaussian 92 can handle non-unit charges -- use the MASSAGE keyword and tell the program which "atoms" are to have their charges modified. If you want point charges (no basis functions), specify the atoms as ghost atoms (symbol Bq), which have no basis functions or electrons but which (unlike dummy atoms) are kept around after the cartesian coordinates have been generated. The program tries to be clever about figuring out the number of electrons (so that setting atomic numbers to 0 is all you need to do for counterpoise calculations) after the charges are changed, so you may need to fiddle with what overall charge you specify for the system to get the correct number of electrons. Mike Frisch ------- From raman@bioc01.uthscsa.edu Mon Mar 15 15:38:39 1993 From: raman@bioc01.uthscsa.edu (C.S.RAMAN) Message-Id: <9303160338.AA09069@bioc01.uthscsa.edu> Subject: Re: AMINO ACIDS by DFT To: sauer@organik.uni-erlangen.de (Wolfgang Sauer) Date: Mon, 15 Mar 1993 21:38:39 -0600 (CST) Here is a recent reference that talks about Dipole moment of Amino Acid side chains derived from Protential an field derived net atomic charges: J. Phys. Chem. [1992] 96, 10276-10284 entitled "Modelling Amino acid side chains. 1. Determination of Net atomic charges from ab initio self consistent filed molecular electrostatic properties. by Scheraga et al. Hope you find this useful! > > From: siroiss@cerca.umontreal.ca (Suzane Sirois) > > I would like to have references for the bond length > > the infrared spectra and the dipole moment of amino acids > > that i can compared with DFT calculation > > > > a more general question to the DFT gurus out there: > how good are DFT calculations on H bonded systems like amino acids or small > (or even larger ?) peptides ? > References, general remarks, practical hints, ... all welcome. > -- C.S.Raman raman@bioc01.uthscsa.edu - Internet UNIX Programming & Administration 70412.2354@compuserve.com - CIS SPARC & SGI Systems raman@hermes.chpc.utexas.edu - CHPC Department of Biochemistry craman@launchpad.unc.edu 7703 Floyd Curl Dr. (210) 567-6623 [Tel] San Antonio, TX 78284-7760 (210) 567-6595 [Fax] ****************************************************************************** If a man's wit be wandering, let him study the Mathematics -Francis Bacon ****************************************************************************** From raman@bioc01.uthscsa.edu Mon Mar 15 15:55:05 1993 From: raman@bioc01.uthscsa.edu (C.S.RAMAN) Message-Id: <9303160355.AA09078@bioc01.uthscsa.edu> Subject: Re: Ribbons---> molscript ?? To: STINCH%IMISIAM.BITNET@OHSTVMA.ACS.OHIO-STATE.EDU (Davide Proserpio) Date: Mon, 15 Mar 1993 21:55:05 -0600 (CST) > Hi, > Please anyone knows of any software to draw protein molecules in Ribbons? > I saw something called Molscript (probably), anyone has some information about > it or about something else? > Thank you for the help. > Davide > There are a few options at your disposal: 1. There is a program called "RIBBON", written by Dr. Mike Carson, now at University of Alabama,Birmingham, which for non-profit use is available for distribution at a nominal charge! One needs to obtain a licence from him in order to use it. He can be reached at "carson@gtx.cmc.uab.edu" or alternatively (205) 934-1983. This program does a wonderful job of representing proteins as ribbons and would be my first choice for doing the same. I recommend it highly. 2. A program available from the Computing Graphics Laboratory, Univ. California, San Francisco, called MIDAS PLUS also does a very good job of representing proteins not only as ribbons, but also in RASTER, BALL & STICK types. In addition it is quite powerful for doing other calculations on proteins, provided you have PDB files for manipulation. The latter can be easily obtained via GOPHER. This program can also be licenced for a small fee. Please contact Eric Pettersen at UCSF if you are interested in obtaining the same. To find out more about MIDAS, refere to "T.E. Ferrin, et. al. [1988] J. Mol. Graphics 6, 13-27. Both of the above mentioned programs would require a SGI workstaion to run. Hope you find these programs useful! -- C.S.Raman raman@bioc01.uthscsa.edu - Internet UNIX Programming & Administration 70412.2354@compuserve.com - CIS SPARC & SGI Systems raman@hermes.chpc.utexas.edu - CHPC Department of Biochemistry craman@launchpad.unc.edu 7703 Floyd Curl Dr. (210) 567-6623 [Tel] San Antonio, TX 78284-7760 (210) 567-6595 [Fax] ****************************************************************************** If a man's wit be wandering, let him study the Mathematics -Francis Bacon ******************************************************************************