From chemistry-request@ccl.net Tue Feb 26 04:17:34 1991 Date: 26 Feb 91 09:56:20 EDT From: UDIM018%FRORS31.BITNET@OHSTVMA.IRCC.OHIO-STATE.EDU Subject: letter to peterson To: chemistry@ccl.net Status: R E. M. EVLETH Dynamique des Interactions Moleculaires Universite Pierre et Marie Curie 4 Place Jussieu, Tour 22, Paris 75005 (1) 43 26 09 82 UDIM018 at FRORS31 (This is a modified copy of a letter sent to Mike Peterson on his request for information on MOPAC 5 and 6) Dear Mike: Per your call for help--we are using a yet unreleased version of AMPAC 214, some of the programming has been done by D. Leotard at Bordeaux, which has full AM1 and PM3 parameterisations. I have done the following check calculations--- SIH--the literature AM1 value is not as you quoted (Organometallics 1987, 6, 1486) but 89.8. This is a doublet species and the Dewar programs use both the half-electron and UHF methods-- His protocol uses always to former (Nesbet) hamiltonian which is not authetically SCF (there are non-zero single replacement matrix elements in the CI expansion). In any case with AMPAC 214 I obtained the following optimized values AM1 parameterisation 1 UHF gives 89.4 kcal SIH = 1.453 2 half-electron 89.8 " SIH = 1.454 The latter is in agreement with Dewar's paper, so your program is defective in some manner, ours is not. I guess. As for HgO, AM1 parameterisation: You give a value of 140.6 MOPAC6 101.6 version 5.0. With AMPAC214 I obtained AM1 102.7 R = 2.062 PM3 message stating that Hg not availalbe in PM3 MNDO 101.6 R = 1.883 I note here that AMPAC 214 states that the Mercury parameters have not yet been published---i.e. better contact DEWAR BITNET = CMBA007 at UTCHPC, Healy is the contact person. As for H2S you give 3.8 and 1.7 on versions 6.0 en 5.0. AMPAC214 gives 1.2 at a SH and HSH optimized parameters of 1.322 and 95.5 degrees respectively--- AMPAC214 gives the sulfur AM1 parameterisation as INORG CHEM 1990 submitted--I have a reprint of that article and it gives the AM1 heat of formation of 1.2. THEREFORE 1. You are working with defective programs, either with wrong parameterizations or defective mountings| 2. I note that QCPE published corrections to MOPAC5.0, last year page 34, for silicon parameters. 3. I have not looked at the AM1 parameter file in MOPAC 5.0 but I do note that certain programs would give a socalled AM1 calculation even if the parameters were NOT IN THE PROGRAM. 4. I have not looked at the AM1 parameter file in MOPAC 5.0 but Early versions of AMPAC did the same thing--in fact there are a SERIES OF SOCALLED AM1 SULFUR CALCULATIONS in the literature which were not done with AM1 sulfur parameters| The earlier programs would grab MNDO parameters and mix'em with AM1 and give a bastard calculations. I refereed a paper a while back which I refused for that reason, recommending the paper be sent to Dewar for review and he commented that it was indeed unfortunate that this was possible. Some people I know (former easter block groups) having put in the literature parameters for AM1 or PM3 in their own home variety programs have not been able to reproduce the literature heats of formation--merely reprogramming with new DATA files is not sufficient. I finally note that AMPAC214 seems relatively error free, is over all better than the MOPAC series. Leotard is a clever programer and has added some optimisation features. Also, I think he has introduced a CIPSI type configuration search in the CI expansion. One does not normally use semiempirical theory at the CI level, but I have done a number of calculations on benzene hypersurfaces in the excited states in which the semiempirical treatment match QUALITATIVELY what is known at the ab initio MCSCF level. It is also known (work of Walter Thiel at Wuperthal) that semiempirical theory can give one a good guess at certain types of transition states (especially if tight). I have a number of examples of ab initio transition state optimizations in which I start with AM1 or PM3 values, modified them slightly if the bond distances appear poor. This includes ion-molecule transition states. I CAN NOT VOUCH FOR WHEN AMPAC 214 will be distributed, DEWAR is the primary author but the following evolution of the program brings one to November of last year. I also not that the following information was not authorized for distribution and if it depleases the authors I give them my apologies-(Je ne veux pas emmerder ces gens). C STARTING POINT: AMPAC VERSION 2.1 ... NOVEMBER 1986 C VECTORIZED FOR CRAY-1S ... JANUARY 1987 C VECTORIZED FOR IBM 3090-VF2 ... FEBRUARY 1988 C BACK TO A 'SCALAR' VAX/IBM PORTABLE VERSION ... DECEMBER 1988 C REVISED FOR IMPLEMENTATION UNDER "UNIX" SYSTEMS ... MAY 1990 C IMPLEMENTATION OF PM3 PARAMETER SET ... JUNE 1990 C IMPLEMENTATION OF "IRC" OPTION, VARIOUS BUGGS FIXED, C IMPROVED C.I. AND RELEVANT GRADIENTS, C LEADING TO THE PRESENT PORTABLE VERSION 2.14 ... NOVEMBER 1990 C IMPLEMENTATION OF MNDOC PARAMETER SET ... NOVEMBER 1990 C C C QUESTION OR COMMENT SHOULD BE DIRECTED TO C EITHER THE M.J.S. DEWAR GROUP, UNIVERSITY OF TEXAS AT AUSTIN C DPT OF CHEMISTRY C AUSTIN TEXAS 78715 C OR DANIEL A. LIOTARD, UNIVERSITE DE BORDEAUX 1 C LABORATOIRE DE PHYSICO-CHIMIE THEORIQUE C 33405 TALENCE FRANCE C GOODLUCK E. M. Evleth --- From chemistry-request@ccl.net Tue Feb 26 10:18:01 1991 Date: Tue, 26 Feb 91 09:41:51 EST From: ra!rgl@cs.psu.EDU Subject: postdoc position wanted To: CHEMISTRY@ccl.net Status: R I have recently finished my Ph.D. with Peter Jurs at Penn State. I am interested in obtaining a postdoc position in computational chemistry in the Philadelphia area (my wife has just accepted a position at ARCO). My thesis research was focussed mainly on data analysis techniques, especially QSAR methods. I have done some molecular modeling and energy calculations (MM2, AMPAC), but this is where I would like to gain more experience. For more information, or for a copy of my resume, I can be reached here by e-mail, or by phone at (814) 865-7402 or (814) 867-2351. Thank you for your attention, Richard G. Lawson --- From chemistry-request@ccl.net Tue Feb 26 12:11:41 1991 Date: Tue, 26 Feb 91 10:47 CST From: AHOLDER@VAX1.UMKC.EDU Subject: Dewar Group To: CHEMISTRY@ccl.net Status: R I response to the message about AMPAC, several things should be cleared up. As a recently former member of the group, I feel at least partially qualified to do this: 1. M.J.S. Dewar is no longer avilable at the University of Texas. He moved to U. of Florida last year and is now permanently located there. 2. Daneil Liotard reprogrammed large portions of all of the AMPAC codes and recent versions incorporate many of his improvements, resulting in what appear to be significant performance gains over older versions of MOPAC. The situation with MOPAC 6.0 remains to be evaluated. 3. Eamonn Healy is no longer available at the BITNET address noted. There is not at present an "official" contact. 4. I appreciate the note about mixed parameter sets. This is a heinous crime against humanity in general. The original PM3 papers published by Stewart featured this type of difficulty. Andy Holder U. of Missouri-Kansas City Department of Chemistry AHOLDER@UMKCVAX1 (Bitnet) --- From chemistry-request@ccl.net Tue Feb 26 13:11:11 1991 Date: Mon, 25 Feb 91 20:06:14 GMT From: jle@stardent.com Status: R I'd like to know how one properly refers to the Protein Data Bank in a paper... In other words, what is the author, site, etc. reference that goes in the footnotes and/or acknowledgement section(s)... Thanks in advance, Joe Leonard Computational Chemist Stardent Computer Inc. jle@Stardent.COM --- From chemistry-request@ccl.net Tue Feb 26 20:00:28 1991 Date: 26 Feb 91 19:46 EST From: JUDITH.SHELLING@ukwang.uky.edu Subject: faculty position To: chemistry@ccl.net Status: R to: chemistry@ccl.net University of Kentucky expects to close applications for Molecular Modeling/Computational Medicinal Chemistry position March 1. A tenure track faculty position at the Assistant Professor level for a computational medicinal chemist is open in the Division of Medicinal Chemistry and Pharmaceutics. This position involves a joint appointment in the UK Center for Computational Sciences, directed by Dr. John Connolly, and includes access to the Center supercomputer. The applicant should be proficie nt in the use of empirical and ab initio methods for the design of new drug entities, and should possess expertise in the application of parallel processin g and quantum chemical programs to macromolecular systems. The person should have an interest in working with medicinal chemist colleagues in applying computational methods to solving problems in drug design, molecular dynamics of proteins and nucleic acids, and biophysical factors in the formulation and quality control of pharmacologically active proteins. It is expected that the successful applicant will establish a vigorous research program funded by extramural support, will participate in teaching in the professional and graduate degree programs, and will guide graduate and post-doctoral students. Applicants should possess a Ph.D. in either medicinal chemistry, biophysics, physical or theoretical chemistry, or a related field. Relevant post-doctoral experience is a requirement. Send resume and the names and addresses of three references to Dr. H. B. Kostenbauder, Director, Division of Medicinal Chemistry and Pharmaceutics, College of Pharmacy, University of Kentucky, Lexington, KY 40536-0082, FAX 606-257-2128, harry.kostenbauder@ukwang.uky.edu ---