From chemistry-request@ccl.net Fri Jul 10 01:51:00 1992 Date: Thu, 9 Jul 92 22:26 EDT From: SML108@PSUVM.PSU.EDU Subject: Re: Looking for a 'PlotAtoms' program for RS/6000 To: dov@menora.weizmann.ac.IL Status: R I have written a free molecular display system which I would be happy to send you if you are willing to sign a licensing agreement which basically states you will not try to distribute it or modify it and call it your own program. It is currently written in gl but will soon support both gl and X windows. Scott From chemistry-request@ccl.net Fri Jul 10 02:33:06 1992 Date: Thu, 09 Jul 92 21:56:06 -0400 From: radomski@mond1.ccrc.uga.EDU Subject: Re: Forcefields archives idea To: chemistry@ccl.net Status: R I like to second Jan's idea very warmly, how often it is that after exhaustive and protracted search through original (and not so original :-)) sources one wakes in a maze of only so slightly different versions? However, talking about parameters or better to say parametri- zations. What is prevailing opinion on a budding approach to develop semi-empirical hamiltonians tailored not for general usage on every kind of a problem, but toward quite specific ones? Especially for molecules for which rigidity inherent in molecular mechanics poses an obstacle, or for which there is lack of experimental data. I am not a mainstream "semi-empiricist", on the contrary - for the last ten months I've dealt almost exclusively with artificial neural networks, BUT this is exactly the reason for my question - neural nets are seemingly the 'mad scientist' dream of ideal tool for any kind of parametrization. And last but not least, do any of you knows about any kind of BBS, e-mail reflector or computer list dedicated solely or mainly to the artificial neural nets? What is the place to look if one is seeking further employment in this area? For the last few months there is particular dearth of any kind of computational-chemistry advertisements in Science, Nature, Ch&E News and such. Am I overlooking some other obvious choice? Regards, Jan P. Radomski Complex Carbohydrate Research Center, University of Georgia, 220 Riverbend Rd., Athens, 30602 GA. Ph. (706) 542-4452 Fax (706) 542-4412 e-mail: radomski@mond1.ccrc.uga.edu From chemistry-request@ccl.net Fri Jul 10 03:20:48 1992 Date: Thu, 9 Jul 1992 22:07 CST From: Andy Holder To: CHEMISTRY@ccl.net Status: R What about a synopsis on the contents of the latest Gordon Conference on Computational Chemistry from an attendee? Andy =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= DR. ANDREW HOLDER Assistant Professor of Computational/Organic Chemistry Department of Chemistry || BITNET Addr: AHOLDER@UMKCVAX1 University of Missouri - Kansas City || Internet Addr: aholder@vax1.umkc.edu Spencer Chemistry, Room 502 || Phone Number: (816) 235-2293 Kansas City, Missouri 64110 || FAX Number: (816) 235-1717 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= From chemistry-request@ccl.net Fri Jul 10 07:50:19 1992 Date: Fri, 10 Jul 92 03:06:45 EDT From: parkesk@rwei00.DNET.roche.COM Subject: Nitrogen geometry in MOPAC To: chemistry@ccl.net Status: R I have recently encountered a number of problems using MOPAC 6 on molecules containing tricoordinate nitrogen which are made pyramidal using both AM1 or PM3 in several cases where X-ray crystalographic data on model compounds shows the nitrogen to be planar. Examples include: * Structures containing the 1-azabicyclo[4,3,0]nonan-2-one skeleton which is incorrectly described as pyramidal, although the analagous [3,3,0] system is pyramidal by both calculation and experiment. * Amidines containing the R-N(1)=C(R)-N(2)HR group which are wrongly predicted to be pyramidal at N(2). I can see that these problems are related to the "peptide linkage problem" but would be interested to know whether anybody else has encountered similar difficulties, or has found, or is working on, a solution. *********************************************************************** Dr Kevin Parkes Roche Products Ltd E-mail: parkesk@rwei00.dnet.roche.com P O Box 8, Welwyn Garden City Herts AL7 3AY Great Britain ************************************************************************ From states@ncbi.nlm.nih.gov Fri Jul 10 15:47:36 1992 Date: Fri, 10 Jul 92 15:47:29 EDT From: states@ncbi.nlm.nih.gov (David States) To: chemistry@ccl.net Subject: Re: Forcefields archives idea Status: R > At the Gordon Conference on Computational Chemistry there was some > informal discussion of making an electronically accessible depot of > Force Field parameters... Jay Ponder, at Washington University St. Lous, and I had given this some thought a couple years ago. Issues that need to be addressed: 1) a database vs. a repository A database implies a standard format data collection maintained by someone who is responsible for its contents. A repository is simply a place where authors can place information and responsability for content and format remain with the depositor. Given the number of molecular modeling packages and flux in the associated file formats, we concluded that it was only feasible to consider a repository. Standard format for the parameters would increase the utility of the collection considerably, but it introduces work of the depositor (or the curator of a DB) and issues about checking and verification. > I would like to initiate some discussion on this with a few comments > and questions of mine to give it a spin. The description of force field > consists of: > 1. description of the mathematical form of the potential energy function, > treatment of 1-3, 1-4, 1-5, etc. non-bonded interactions, special > cases (e.g. H-bonds), combination rules for vdW, etc. > 2. description of atom/bond types. This may be verbal, but has to be > precise. The information on how these are assigned is not always > trivial (e.g., precedence rules), but is always important. > 3. list of parameters in some kind of a table which relates to 1. and 2. > (with units !), This is implicitly a many-body point (atom) mass description. Is it adequate for all cases? How about Brownian dynamics? stochasitic boundary conditions? semi-empirical electonic structure caluclations? polarization? One could invent a "force field language" and ask depositors to code their potential into it, but this is not a trivial task (either formulating the language or coding a complex PE function in it). Alternatively, one could ask for existing source code deposits, but if someone is not willing to make code available through existing channels (QCPE, ftp, etc.) why would they do it here? > 4. comments on derivation of the FF (experimental, ab initio, what data, > what basis sets, CFF way or "manual" fitting. A useful request and not amenable to much standardization. 2) why should authors deposit? Not clear to me. Maybe it would be convenient to have a central clearing house, but for the individual author/developer it seems like a local ftp server would be a preferable solution. You keep control, know who has been accessing it, updates are easier, and people who download know where to direct their questions. > And, what about commercially derived FF --- are they proprietary, and > if so, should the papers using them be published in reviewed journals > or they fit the scope of the famous "Journal of Irreproducible > Results"? Please, no flames, this is not a clear issue, since deriving > force fields is costly but at the same time it is needed. Can you > copyright or patent the number? Not now, and FF developers do not have > much legal defence beside a contract(license). People in software > companies, please speak up ! To my view this is the one group that would benefit most from a central repository. By publishing the FF and allowing people to test it, they would gain acceptance. They make money selling programs, not PE functions. If you have decided that X companies PE func is best, you are much more likely to invest in the companies software suite even if you can hack their PE func into some other software. Especially true for commercial users where support is more of an issue. You can not copyright a number, but you can copyright a particular form a data set and copy/use restrictions are well established (ex. video rentals). All of the commercial parameter sets that I have seen include copoyright notices quite explicitly. So a PE func could be made available with restriction that would make it illegal to copy it and sell it as part of another product. > So in short, we need to agree on some guidelines if we want to do it > (the format of FF description and parameters), the way FF's are named > and referred to (we all know how fluid the parameters of a given force > field are sometimes. It would be nice to have some REFNO kind thing > which refers to a given version of a force field located precisely > in time and space...). References to papers where FF was applied? > References to papers which list parameters for the first time? If you seriously want to see a repository happen, a major journal publisher (ACS) could insist that modeling papers provide a means by which readers could obtain their PE function. Programs which were based on PE functions not available through ftp or a repository would rapidly fall out of favor and there would be a real incentive for developers to deposit their PE functions. It sounds Draconian, but how can an interested reader really evaluate a modeling paper without knowing the PE function? Certainly this approach has worked dramatically in making molecular sequence and structure data more available. David States National Center for Biotechnology Information / National Library of Medicine From chemistry-request@ccl.net Fri Jul 10 17:14:10 1992 Date: Fri, 10 Jul 92 10:36:00 +0200 From: oosten@rulhm1.LeidenUniv.NL (A.B. van Oosten) Subject: H2+ To: CHEMISTRY@ccl.net Status: R I need hyperfine data on the H2+ (HD+,D2+) molecular ion, either experimental or theoretical. Who can provide me with references to relevant literature ? I need the data for the analysis of the EPR spectrum of H2 in crystalline silicon. Thanks Tony van Oosten oosten@rulhm1.leidenuniv.nl From chemistry-request@ccl.net Fri Jul 10 18:37:36 1992 Date: Fri, 10 Jul 1992 9:34:21 -0400 (EDT) From: HERMAN@ULNA.BWH.HARVARD.EDU To: chemistry@ccl.net Status: R A few thoughts on the standardization of force force fields: 1) A list of standard atom types should evolve, which parallels real, unique atom types. That is, a ketone carbonyl should be different from an acid carbonyl, etc... This list would be maintained under IUPAC auspices. 2) A list of standard parameter types should evolve. The rub here is that all force fields use different parameter types. OK. It is up to the user whether he/she will extract one parameter, from one force field, and do whatever verification is necessary to establish the appropriatness of using that parameter with his/her force field. 3) Parameters should be PUBLIC DOMAIN. Hiding parameters in proprietary software adds to the characterization of that software as a black box. Publications should include all parameters used for the systems looked at, and should include an example of some type of verification of the parameters. 4) Parameters should be readily accessible through the Internet. 5) Parameters should always be stored by software in an easily edited, ASCII text file. Preferably, the software should include a dedicated parameter editor. Parameters should always be included with output files. 4) A brief, standard notation describing the force field and its parameters should evolve, in order to keep wordiness in the literature to a minimum. Lee H. From liu@chem.nwu.edu Fri Jul 10 18:33:06 1992 Date: Fri, 10 Jul 92 17:30:59 CDT From: liu@chem.nwu.edu (Shengzhong Liu) To: chemistry-request@ccl.net, dov@menora.weizmann.ac.il Subject: Re: Looking for a 'PlotAtoms' program for RS/6000 Status: R I agree to all those conditions. Could you please send me a copy? Thank you. From chemistry-request@ccl.net Fri Jul 10 20:02:53 1992 Date: 10 Jul 1992 10:35:55 -0400 (EDT) From: "DR. DOUGLAS A. SMITH, UNIVERSITY OF TOLEDO" Subject: Re: Nitrogen geometry in MOPAC To: parkesk@rwei00.DNET.roche.COM Status: R Tricoordinate nitrogen (i.e. amine nitrogen) geometries are not well produced by the semiempirical methods available in MOPAC, i.e. MNDO, AM1 or PM3. We recently published a study comparing these three methods to ab initio and experimental results for a series of 15 aromatic amines and two DNA intercalating compounds, 0-AMSA and m-AMSA. AM1 did the best job overall, but not particularly well. The major reason for this (J. Stewart/A. Holder correct me if I am too far off base) is that nitrogen was parameterized (as were all elements) using a broad range of structurally and electronically variant compounds. The nitrogen problem seems to arise particularly from the inclusion of NO2 compounds. Ref: Smith, D. A., Ulmer, C. W. II, Gilbert, M. J. J. Comp. Chem. 1992, 13, 640. reprints can be requested by email or USmail from me. Doug Smith Assistant Professor of Chemistry The University of Toledo Toledo, OH 43606-3390 voice 419-537-2116 fax 419-537-4033 email dsmith@uoft02.utoledo.edu From chemistry-request@ccl.net Fri Jul 10 20:37:06 1992 Date: Fri, 10 Jul 92 16:18:01 -0400 From: "PATHOLOGY & LAB. MEDICINE" Subject: Neural network mailing list To: chemistry@ccl.net Status: R Jan Radomski recently requested information about bulletin boards or e-mail lists devoted to neural networks. One newsletter that contains a fair amount of traffic in this area is "Neuron Digest," which is distributed by the Department of Psychology at the University of Pennsylvania. Requests for addition to the distribution list can be sent to neuron-request@cattell.psych.upenn.edu. FTP archives can be accessed anonymously at cattell.psych.upenn.edu in /pub/Neuron-Digest. Gregg Wells Department of Pathology University of Pennsylvania Philadelphia, PA USA e-mail: pathology@a1.mscf.upenn.edu From chemistry-request@ccl.net Fri Jul 10 21:10:00 1992 Date: 10 Jul 92 09:19:00 PST From: MAX VASQUEZ Subject: Information about Tetrahedron Computer Methodology To: chemistry Status: R Hello, Does anyone have information (subscription, address, phone number, etc) about the +/- new journal 'Tetrahedron Computer Methodology'? Answer directly to me or to the net whatever is easier. Thanks in advance, Max Vasquez From chemistry-request@ccl.net Fri Jul 10 22:58:10 1992 Date: Fri, 10 Jul 92 13:48:34 -0500 From: lwalsh@nemo.life.uiuc.EDU (Laura L. Walsh) Subject: Z-matrix -> Structure To: chemistry@ccl.net Status: R This is transmitted for a friend: >Hello, > >I am creating z-matrices for use as input to Gaussian and I would like to >visually check them before running Gaussian so as not to waste cpu time. Is >there any visualizer that will accept data in the format: > >N N1 distance N2 angle N3 dihedral > >and display a structure? We do not have AVS at this site. > >Cynthia Gibas cgibas@wraightc3.life.uiuc.edu >1201 W. Gregory Ave. Rm. 162 4223 Beckman Institute >Urbana, IL 61801 333-8725 244-2894 For those who wish to reply directly to the list, I will forward any replies to her. Laura Lynn Walsh (lwalsh@nemo.life.uiuc.edu) (217) 244-2894 Beckman Inst., Univ. of IL * 405 N. Mathews Ave. * Urbana, IL 61801, USA From chemistry-request@ccl.net Fri Jul 10 23:00:35 1992 Date: Fri, 10 Jul 92 12:40:53 -0400 From: mckelvey@Kodak.COM To: jkl@Kodak.COM, osc@Kodak.COM Status: R Comment on request for synopsis of the recent Comp Chem Gordon Conference: Any such synopsis MUST BE CLEARED BY THE GRC OFFICE. Reproduction of any ideas of the speakers, and discussions ARE EXPRESSLY FORBIDDEN BY THE GRC RULES AGREED TO IN ADVANCE BY EACH ATTENDEE AS A CONDITION FOR ATTENDING. I will contact the GRC office and find out what can be said, and will pass it on. Sincerely yours, John McKelvey, Chair 1992 Comp Chem Gordon Conference Research Labs Eastman Kodak Co. Rochester, NY 14650-2216 Phone 716-477-3335 From chemistry-request@ccl.net Fri Jul 10 23:31:14 1992 Date: Fri, 10 Jul 1992 15:53 CST From: Andy Holder Subject: Gordon Conferences and Ritual Suicide To: CHEMISTRY@ccl.net Status: R It appears that I got to learn something today. That little something was that the goings-on at Gordon Conferences are not discussed outside of those gatherings. My mistake for asking for a synopsis. My apologies to one and all. Please know that it was sheer ignorance that sparked the question, not malice afore- thought. Andy p.s. Sorry for taking up the bandwidth. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= DR. ANDREW HOLDER Assistant Professor of Computational/Organic Chemistry Department of Chemistry || BITNET Addr: AHOLDER@UMKCVAX1 University of Missouri - Kansas City || Internet Addr: aholder@vax1.umkc.edu Spencer Chemistry, Room 502 || Phone Number: (816) 235-2293 Kansas City, Missouri 64110 || FAX Number: (816) 235-1717 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= From chemistry-request@ccl.net Sat Jul 11 00:14:47 1992 Date: Fri, 10 Jul 1992 18:19:47 -0400 (EDT) From: MAALOUF@RISVAX.ROWLAND.ORG Subject: matrix solvers To: chemistry@ccl.net Status: R Howdy, I'm looking for a QR decomposition or similar matrix solver written in MPL or MPFORTRAN for a DEC MasPar. Does anyone have such code that they're willing to share? Thanks George Maalouf maalouf@risvax.rowland.org Harvard University - biophysics and The Rowland Institute for Science From chemistry-request@ccl.net Sat Jul 11 01:15:46 1992 Date: Fri, 10 Jul 92 18:36:32 -0400 From: radomski@mond1.ccrc.uga.EDU Subject: Re: Nitrogen geometry in MOPAC To: parkesk@rwei00.dnet.roche.COM Status: R Kevin Parkes wrote: >I have recently encountered a number of problems using MOPAC 6 on >molecules containing tricoordinate nitrogen which are made pyramidal >using both AM1 or PM3 in several cases where X-ray crystalographic >data on model compounds shows the nitrogen to be planar. Yes!!! We got exactly the same problem while trying to model N-linked glicosides here at the CCRC. It doesn't matter if one is using MOPAC, AMPAC or G90, and it holds true for old MNDO, AM1 and PM3 (the worst offender). Regards Jan. Jan P. Radomski Complex Carbohydrate Research Center, University of Georgia, 220 Riverbend Rd., Athens, 30602 GA. Ph. (706) 542-4452 Fax (706) 542-4412 e-mail: radomski@mond1.ccrc.uga.edu From chemistry-request@ccl.net Sat Jul 11 02:25:56 1992 Date: Fri, 10 Jul 92 20:13:12 -0400 From: radomski@mond1.ccrc.uga.EDU Subject: Re: H2+ To: chemistry@ccl.net Status: R Tony, contact Bogumil Jeziorski , the Warsaw group of Kolos, Jeziorski & Piela is top notch in accuracy. Another choice is Andrzej Sadlej at Lund, Sweden but I have not his e-mail, or Krzysztof Szalewicz at Univ of Delavare. Regards, Jan. Jan P. Radomski Complex Carbohydrate Research Center, University of Georgia, 220 Riverbend Rd., Athens, 30602 GA. Ph. (706) 542-4452 Fax (706) 542-4412 e-mail: radomski@mond1.ccrc.uga.edu From chemistry-request@ccl.net Sat Jul 11 02:34:37 1992 Date: Fri, 10 Jul 92 21:43:54 -0400 From: radomski@mond1.ccrc.uga.EDU Subject: Re: Neural network mailing list To: "PATHOLOGY & LAB. MEDICINE" Status: R Dear Netters, many warm thanks to ALL of you who answered to my yesterday's questions, the response is quite overwhelming. I've got many additional queries though, about neural nets in general, their performance, benchmarks and such. I'm still trying to cope with them. Maybe the best idea would be to summarise for this list as well? Esp. that besides the "Neuron Digest" Gregg Wells mentions there are many other BBS', newsletters and e-mail lists dealing with ANN. I'm just trying to access them all, so bear with me please. Thanx again, Jan Jan P. Radomski Complex Carbohydrate Research Center, University of Georgia, 220 Riverbend Rd., Athens, 30602 GA. Ph. (706) 542-4452 Fax (706) 542-4412 e-mail: radomski@mond1.ccrc.uga.edu