From yverband@is1.bfu.vub.ac.be Wed Jun 30 02:45:43 1993 From: yverband@is1.bfu.vub.ac.be (Verbandt Yves) Message-Id: <9306300645.AA05451@is2e.vub.ac.be> Subject: overview biphenyl data To: CHEMISTRY@ccl.net Date: Wed, 30 Jun 93 8:45:43 MDT Dear Netters, Here is the overview of the responses I have got on my posting about the geometrical data on biphenyl. No answers on my posting about the optical data on biphenyl have reached me yet. Thank you all, Yves. ================================================================ Yves Verbandt tel. 32-2-641.34.54 Brussels Free University fax. 32-2-641.34.50 Applied Physics Departement email yverband@vub.ac.be Pleinlaan 2 B-1050 Brussel BELGIUM ================================================================ >From: kudchadk@ecn.purdue.edu (Samir Kudchadkar) Hi Yves, This may not be what u want, but I happen to be aware of a Molecular Dynamics Simulation of solid diphenyl.The reference is " MD simulation of solid biphenyl " by Baranyai and Welberry, Molecular Physics, 1991, Vol. 73, No. 6, 1317- 1334. It has a lot of references on expt. studies on the twist angle in the solid and structural phase transitions in the solid phase. Hope this helps. Good luck, - Samir Kudchadkar Graduate Student Dept. of Chemical Engg. PURDUE UNIVERSITY. >From: Suse Broyde akiyama and watanabe j phys chem 1986 90 1752-1755 From sauer@organik.uni-erlangen.de Wed Jun 30 03:39:41 1993 From: Wolfgang Sauer Message-Id: <9306300711.AA19933@derioc1.organik.uni-erlangen.de> Subject: Re: helix construct To: JSCHELDE@steevast.rug.ac.be, CHEMISTRY@ccl.net (CompChem MailExploder) Date: Wed, 30 Jun 93 9:11:52 METDST > > Hi all, > > Can somebody tell me the best way to constuct a helix(polyalanine,..) > to use then in M.D. simulations. > Thanks. > Dear Jean-Luc, I can think of several ways, depending on how often and how general you need it: 1) Use of a GUI of some other program (e.g. HyperChem) that lets you create peptides by adding residues in selected conformations and then reformat the coordinates to whatever your proggram needs. Just select "alpha" and click the "Ala" button as often as you like ... 2) Use of internal coordinates: it is much easier to write a Zmatrix for a problem like this and then use one of the Z->cart converters that were a topic on this list sometime ago. 3) Write a program to do it for you. This is probably the most work, but if you need many helices with varying n,d,... values this effort might be worthwhile. For a ref. see R.E.Koeppe, M.Kimura: Biopolymers _23_, 23-38 (1984) and refs cited therein. There the repeating unit is two residues but this can easily be changed to whatever you need. 4) If you really need just Ala and your "poly" is not too long, you can get xyz's from me. I used 2), but since I did qm calcs and the machine was not too big, I've got only up to n=13. But if you have a way of glueing pieces of that size together, you are welcome. Or I can create a chain using HyperChem for you. Hope this helps. All the best, Wolfgang. -- +======================================+=====================================+ ! Please note the change of all German postcodes from 1 July 1993 ! +======================================+=====================================+ | Wolfgang Sauer | | | Institut fuer Organische Chemie I | "You have to be three standard | | Henkestr. 42, D-91054 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 rwoods@biop.ox.ac.uk Wed Jun 30 13:11:55 1993 Date: Wed, 30 Jun 1993 12:11:55 +0100 From: rwoods@biop.ox.ac.uk To: wipke@SECS.UCSC.EDU Message-Id: <0096ECC8.368B1FA0.9286@biop.ox.ac.uk> Subject: RE: Intellectual Property Three cheers to Todd Wipke for his timely comments regarding scientific ethics in the field of computational chemistry. I don't believe that computational chemists are any more prone to unethical publishing procedures than any other scientists. However, the amount of detail that belongs in a computational paper, to ensure reproducibility, is probably greater than in most other fields. It takes a very conscious effort to include all of the experimental details, an effort that some computational chemists fail to make. It may be that the authors "assume" everyone knows the details, or they may be careless or lazy, or it may be a deliberate attempt to deter others in the same area of research. I think the only way to improve the situation is through very meticulous and conscientious refereeing. As a former synthetic chemist I am well aware of papers published by extremely well known synthetic "leaders" that do not contain sufficient detail to reproduce the work. These papers may have been accepted because of the high profile of the authors. I think we have to be very careful to avoid this situation if, as a comparatively new science, we are to avoid losing respectability. Just my thoughts, Rob Woods Robert J. Woods, Ph.D. Glycobiology Institute Department of Biochemistry University of Oxford South Parks Road Oxford, OX1 3QU From mercie@cumc.cornell.edu Wed Jun 30 03:54:42 1993 Date: Wed, 30 Jun 1993 07:54:42 -0400 (EDT) From: Gustavo Mercier Subject: Re: Silicon Graphics To: SC18000 Message-Id: On Tue, 29 Jun 1993, SC18000 wrote: > Hi Netters, > Could someone quickly give me the address and phone number of Silico > n Graphics, Inc? I need to speak to people about prices etc. > Thanx, > Ken Fountain/ sc18@nemomus > > SGI's number is 1-800-800-SGI1 (the last one is "number 1") gus mercie@cumc.cornell.edu From ZAHAREVITZ@dtpvx2.ncifcrf.gov Sun Jun 30 03:39:00 1993 Message-Id: <199306301302.AA27320@oscsunb.ccl.net> Date: 30 Jun 93 08:39:00 EST From: "Zaharevitz, Daniel" Subject: RE: LogP calculations? To: "chemistry" Kestutis Sliupas asks: > Hi: > I was wondering if there was any public domain code > available for calculating the octanol / water partition co- > efficient? This would be for small molecules, less than say mw350. > Archie returned nothing useful searching for 'logp'. Which reminds me I haven't summarized the responses I got to a similar question a few weeks ago. A number of people ( Harold Cox, George Famini, Gregory Durst, Yvonne Martin, Martin Sage ) mentioned CLOGP which came out of work by Leo and Hansch at Pamona and is now distributed by Daylight. Its said to be fast with lots of other useful information, but expensive ($10-15K) For information, contact: Yosi Taitz or David Leo Daylight Chemical Information Systems 3951 Claremont St., Irvine CA 92714 714 586 6927 FAX 714 476 0654 Robert Funchness and Mario Blanco mentioned Polaris from Molecular Simulations. It is based on the work of Warshel: Warshel, A. and Aqvist, J., Ann. Rev. Biophys. Chem. 20, 267 (1991) Luzhkov, V. and Warshel, A., Computers in Chemistry, Feb. (1992) Dr. Blanco estimated that calculating log P for 100,000 compounds would take about 5 days on an SGI Indigo. Gregory Durst has used a program called HINT! which is available for a relatively small fee from Glen Kellogg Dept of Medicinal Chemistry Virginia Commonwealth Universiy Richmond, Virginia 23298 phone: 804/786-8490 Martin Sage says there is a program called LOGP for PC's that is available for $600 by contacting Dr. Phil Howard (315)426-3350. Martin Norin and David Ricketts listed a number of useful references Rekker, R.F., de Kort, H.M. (1979). "The hydrophobic fragmental constants; an extension to 1000 data point set". Eur. J. Med. Chem.-CHIMICA THERAPUTICA. 14, pp 479-488. Leo & Hansch - J.Med Chem 1976 19 p611 Kellogg & Abraham - J Mol Graphics 1992 10 p212 Dunn, Koehler & Grigoras - J Med Chem 1987 30 pp1121-1126 Bodor Gabanyi & Wong - J Am Chem Soc 1989 111 pp3783-3786 Lewis J Comp Chem 1989 10(2) pp145-151 Ghose & Crippen - J Comp Chem 1986 7(4) pp565-577 Essex, Reynolds & Richards - Chem Comm 1989 6 ??? Many thanks to all for the useful information. Daniel Zaharevitz zaharevitz@dtpvx2.ncifcrf.gov Information Technology Branch Developmental Therapeutics Program National Cancer Institute From nealn@candyman.esd.sgi.com Wed Jun 30 02:29:51 1993 Date: Wed, 30 Jun 93 09:29:51 -0700 From: nealn@candyman.esd.sgi.com (Neal Neuburger) Message-Id: <9306301629.AA26416@candyman.esd.sgi.com> To: CHEMISTRY@ccl.net Subject: Re: Silicon Graphics >> Could someone quickly give me the address and phone number of Silico >> n Graphics, Inc? I need to speak to people about prices etc. >1-800-800-SGI1 or 1-800-800-7441 The above number is for SGI express. The main number is 800-800-4SGI or 800-800-4744 Neal From jle@world.std.com Wed Jun 30 09:45:21 1993 Date: Wed, 30 Jun 1993 13:45:21 -0400 From: jle@world.std.com (Joe M Leonard) Message-Id: <199306301745.AA11430@world.std.com> To: chemistry@ccl.net Subject: A clarification... Folks, I wanted to follow Graham's parenthetical comment re: MM2 and how the program contains "odds and ends" that aren't published. Our efforts with the MM2 and MM3 forcefields have been based on the published literature (along with various sessions spent rationalizing differences between our's and published geom's and energies). Also, we provide versions of FORCEFIELDS not PROGRAMS (as I imagine Hyperchem and others do as well). Granted, there are things in the "original" works that aren't obvious so far, but then they've not come up when working with the published information. SPARTAN V3.0 now has molecular mechanics capabilities, and it's been a tedious but important process verifying all those published structures, etc. This is the only way we can claim that our implementations of the FF's are true. The creators of the MM2/MM3 forcefields (as well as the programs) have two "products" - one scientific, which is a useful, relatively complete mechanics forcefield and the other, which is a computer code. They are interested in areas that I, personnally, am not, but that's probably true in reverse as well. There's probably a whole range of software engineering (read, unpublishable) issues that I'm interested in that are irrelevant to academic developers. Joe Leonard jle@world.std.com From CHEM8@VAXA.YORK.AC.UK Wed Jun 30 14:35:50 1993 Message-Id: <199306301806.AA02327@oscsunb.ccl.net> Date: Wed, 30 Jun 93 19:07 BST From: "John Waite, Tel 1-7238958, N.H.R.F., Vas. Konstantinou 48, Athens 116-35" To: CHEMISTRY Subject: Contact for Dr. F.B.van Duijneveldt? Does anyone have an address and/or phone and/or fax for Dr. F.B. van Duijneveldt, previously of IBM, San Jose, USA, now believed to be in Holland? John Waite chem8@vaxa.york.ac.uk From CHEM8@VAXA.YORK.AC.UK Wed Jun 30 15:04:26 1993 Message-Id: <199306301815.AA02499@oscsunb.ccl.net> Date: Wed, 30 Jun 93 19:16 BST From: "John Waite, Tel 1-7238958, N.H.R.F., Vas. Konstantinou 48, Athens 116-35" To: CHEMISTRY Subject: Public domain computer games? Hi netters, Does anyone know where some computer games, to run under DOS, can be obtained from, in the public domain? Hopefully, John Waite chem8@vaxa.york.ac.uk From mercie@cumc.cornell.edu Wed Jun 30 09:58:33 1993 Date: Wed, 30 Jun 1993 13:58:33 -0400 From: mercie@cumc.cornell.edu (Gustavo Mercier) Message-Id: <199306301758.AA21208@cumc.cornell.edu> To: chemistry@ccl.net Subject: reproduciblity, codes, and disclosure Hi Netters! There has been significant discussion about reproducing computational results from the literature, code copyrights, and distribution of source codes. Below I describe a possible "work around" that may satisfy developers and users. Most individuals that wish to receive source code probably fall in three categories: 1) code scavengers, 2) code "extenders". The scavenger likes to use subroutines available in other codes to ease the construction of his own. In computational chemistry this is probably done very often (i.e. swaping two electron integral codes etc.) without proper recognition of the original author. Developers frown upon this practice and with good reasons. The code "extender", as I define him/her, simply wishes to add modules to an existing code without altering it (except for the proper interface) or use new parameters within the coded model. It is in the interest of developers and users to encourage this practice with proper citations. But how are developers going to do this without distributing their code? Here is my work around! Consider the case of MATHEMATICA from Wolfram Research (I am certainly not promoting their product!). The code is not distributed, but two avenues exist to "extend" the code. One, is its built-in programming laguange. The seconde is its standard interface called MATHLINK. Building a script language is alredy being done for some commercial packages in molecular modeling (i.e. Hyperchem etc), although many are limited to repeating their Graphical Interface functions within a command line mode of operation. MATHLINK is more interesting! It allows a user to write any code in any language (although C and Fortran are easier to implement) and interface it to MATHEMATICA. A user can access from his program any module from MATHEMATICA that developers have allowed access. Also, the converse is true! As an example, Spyglass's Transform v. 3.0 for Unix systems is interfaced with MATHEMATICA in this way. By doing this the developers of Spyglass can concentrate in coding for image presentation rather than the mathematical transformation of images since this can be handled by MATHEMATICA. Moreover, users have all the feature of MATHEMATICA to their disposal. It is possible for a developer to generate such "linking programs" so that the code "extenders" may expand the functionalities of a program without messing up or distributing their code. If they don't want to build their own "linking" standard and languange, they can use MATHLINK. For those, that wish to use their own parameters, an implementation that reads tables of parameters rather than "hardwiring" them into the code is a possiblity. gus mercie@cumc.cornell.edu ps. I have found in my work that programming in Mathematica can help me concetrate in the structure of the code. This makes subsequent programming in C or Fortran much easier. From mercie@cumc.cornell.edu Wed Jun 30 09:58:54 1993 Date: Wed, 30 Jun 1993 13:58:54 -0400 (EDT) From: Gustavo Mercier Subject: reproducibility, codes, and distribution To: chemistry@ccl.net Message-Id: Hi Netters! There has been significant discussion about reproducing computational results from the literature, code copyrights, and distribution of source codes. Below I describe a possible "work around" that may satisfy developers and users. Most individuals that wish to receive source code probably fall in three categories: 1) code scavengers, 2) code "extenders". The scavenger likes to use subroutines available in other codes to ease the construction of his own. In computational chemistry this is probably done very often (i.e. swaping two electron integral codes etc.) without proper recognition of the original author. Developers frown upon this practice and with good reasons. The code "extender", as I define him/her, simply wishes to add modules to an existing code without altering it (except for the proper interface) or use new parameters within the coded model. It is in the interest of developers and users to encourage this practice with proper citations. But how are developers going to do this without distributing their code? Here is my work around! Consider the case of MATHEMATICA from Wolfram Research (I am certainly not promoting their product!). The code is not distributed, but two avenues exist to "extend" the code. One, is its built-in programming laguange. The seconde is its standard interface called MATHLINK. Building a script language is alredy being done for some commercial packages in molecular modeling (i.e. Hyperchem etc), although many are limited to repeating their Graphical Interface functions within a command line mode of operation. MATHLINK is more interesting! It allows a user to write any code in any language (although C and Fortran are easier to implement) and interface it to MATHEMATICA. A user can access from his program any module from MATHEMATICA that developers have allowed access. Also, the converse is true! As an example, Spyglass's Transform v. 3.0 for Unix systems is interfaced with MATHEMATICA in this way. By doing this the developers of Spyglass can concentrate in coding for image presentation rather than the mathematical transformation of images since this can be handled by MATHEMATICA. Moreover, users have all the feature of MATHEMATICA to their disposal. It is possible for a developer to generate such "linking programs" so that the code "extenders" may expand the functionalities of a program without messing up or distributing their code. If they don't want to build their own "linking" standard and languange, they can use MATHLINK. For those, that wish to use their own parameters, an implementation that reads tables of parameters rather than "hardwiring" them into the code is a possiblity. gus mercie@cumc.cornell.edu ps. I have found in my work that programming in Mathematica can help me concetrate in the structure of the code. This makes subsequent programming in C or Fortran much easier. From mail Wed Jun 30 03:55:30 1993 Date: Tue, 29 Jun 1993 18:00:22 -0400 From: hyper!hurst (Graham Hurst) Message-Id: <9306292200.AA11850@hyper.hyper.com> To: chemistry@ccl.net Subject: Re: SAM1 reference and assorted comments following Thanks to Andy Holder for his clarification and to all the others who have added to this thread. Obviously I am not alone in my passion for reproducibility! I agree with Henry Rzepa that referees are crucial in establishing and upholding the standard. Of course it is unreasonable to expect a referee to try to reproduce every result in a paper that he or she has been sent to review, but assesing reproducibility is, or should be, one of the most important aspects of a referee's job. At the risk of echoing Alain St-Amant, David States, and Mark Thompson, I feel it is important to distinguish "science" from "code". The former includes everything a researcher needs to know to reproduce numbers, eg. equations, fundamental constants, fitted parameters. The latter implements the former and will also have data structures, control structures, user interface, programming techniques, performance tuning, etc. My view is that the "science" *must* be fully and publicly disclosed if results are going to be published, but "code" need not be. I strongly agree with Mark Thompson that if the *method* is proprietary, the results should be as well! A *program* can be proprietary, but if the *method* is published then the results can be as well. I disagree with Doug Smith's assertion that this view is utopic and that by this reasoning results from proprietary programs like Gaussian 92 should not be published. The example David States gave of enforcing full disclosure (w.r.t. data rather than method) in the the case of protein structures and sequence data shows that scientific standards can be upheld. In the case of Gaussian 92, the *science* has been published so even though its implementation is proprietary, results from it can be (and are!) independently verified and thus its results are publishable. Like Mark Thompson, Hypercube has implemented several methods from scratch by reading the papers and verifying the equations. To answer Mark's question about experiences in implementing MNDO-family methods, I didn't do that work but have been told by Yufei Guo (who did) that other than typos, units and the absence of a few integral equations, he can't recall anything missing from the papers. However he did have access to public domain MOPAC source, so he might have found there the unpublished "special corrections" that Andy Holder has referred to. Doug Smith pointed out that the problem of reproducibity is notorious for molecular mechanics results. Having implemented AMBER, CHARMM and OPLS from scratch and transformed public domain MM2 into C, I can say that the standards for reproducibility are uneven. Except for MM2, the functional forms are well-described, but descriptions of charges, assignment of atom types and specification of improper torsions are often omitted. For CHARMM, the oft-referenced PARM19 set has never actually been published (only non-bonded parameters are in Walter Reiher's thesis, and that is usually given as the reference). For MM2, as Lou Allinger has admitted, the only complete specification is the program itself (I suspect Joe Leonard has reached this conclusion independently!). Doug Smith brought up the problem of specifying "generated" parameters - in this case I believe it is sufficient for reproducibility to describe the algorithm and data used to generate these parameters. Andy Holder also wrote: > It should be noted, however, that some interests are not scientific, but > competitive. This aspect of the situation seems to be of the most > concern to Dr. Hurst and Hypercube. This company includes a number > of Michael Dewar's previously released methods in their software. Ouch! As I said to Andy Holder when he called me immediately after my email, of course I have a vested interest in the full public disclosure of scientific methods, but I believe it is a vested interest and important issue for all scientists! My work experience has perhaps strengthened this conviction but commercial competition is not "of most concern" to me! (Many readers of this reflector and the usenet sci.chem group may recall me venting on the subject before, as I fear many scientific acquaintances will as well!) I should clarify that in my earlier email (and in this or any other), I am representing my own personal views unless otherwise stated. Cheers, Graham ------------ Graham Hurst Hypercube Inc, 7-419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 internet: hurst@hyper.com From cramer@staff.tc.umn.edu Wed Jun 30 07:44:42 1993 Message-Id: <0012c31d10b005070@staff.tc.umn.edu> From: "Christopher J Cramer-1" Subject: Just for argument's sake . . . To: chemistry@ccl.net Date: Wed, 30 Jun 93 12:44:42 CDT Netters, This whole breast-beating thing is getting a bit out of control, in my opinion. Several thoughts: I object strongly to the contention that computational chemists as a group are somehow less ethical than other chemists. I suspect that the group of fourteen people whom the NIH ORI recently placed into the Federal Register as guilty of scientific misconduct did not include any computational chemists -- hell, there's not enough money in our field to make being unethical worth the trouble (yet)! Certainly, I would like to see ALL scientists maintain the highest standard of ethicality. I resist, however, the analysis that implies it is a given field which leads to misconduct, as opposed to a given individual. As for reproducibility, etc. Every scientific instrument in the world can trace back to some researcher who built the first one. If I'm one of those people, am I forbidden to publish until I allow a referee to come use my machine or build his own?? Nonsense. If it's spectacular science, then people WILL build duplicate machines, and probably calibrate themselves on my published work, and I'd be a fool to have lied about it. If it's such boring science that no one will ever attempt its reproduction, then who cares if it's a lie slipped into the literature? I'll never be cited and my publication won't do me much good in the long run. Mind you, I'd like to believe VERY, VERY few scientists would ever even consider this kind of publication. What I'd like to emphasize is that one is forced to trust one's colleagues to a large extent. A few high-profile misconduct cases are hardly reason to totally lose one's faith in decent human nature. It seems an unfortunate byproduct of the 80's that we all suddenly equate even the possibility of impropriety with it necessarily having occured. As for disclosure of code, etc., this sounds like a lot of whining to me. If you don't like having to pay a lot of money for a code, don't buy it -- code your own. Think that's a lot of work? Of course it is! Why do you think they're charging you money for it?? Many people seem to object to seeing anything in print involving SAM1 because the details of the method, including parameters, are not available. This is bizarre. I can think of twenty chemical reactions, many of tremendous economic importance, where it is not at all clear what the mechanisms are, what the catalysts are, whether you have solutions or microemulsions, etc. but the results are of significant importance. If SAM1 had dropped out of the sky in an impregnable black box that gave answers to questions, and it seemed like those answers were usually correct, then I would use the box to try to gain further insight into chemistry. Sure I'd want to know why it was correct, and I'd beg Andy Holder to help me figure it out (and I note that poor Andy, whose private e-mail is evidently fair game for international mail-explosion, seems to be saying that he is going to disclose everything eventually -- I'm not certain he's under any obligation to do so), but in the meantime, I'd judge the worth of the model by how well it predicted reality. Since when have journals been allowed only to publish papers where no questions remain to be resolved? I point out, for those of you who are going to find my commentary inflammatory, that the authors of AMSOL, of which I am one, have committed themselves to making every aspect of the program and theory available to all, for nothing but the costs that QCPE and journals charge. We've adopted this philosophy because we think it's how science should be done. However, we do this because we can afford to, at least temporarily. In the darkest night of no-funding, who knows what we might be forced to do in order to pay graduate students, postdocs, buy computers, etc.? Who knows whether [insert your favorite commercial code] would be as powerful as it is now were it never to have gone commercial? I am delighted to see that public pressure has caused more disclosure of force field parameters, for instance, from proprietary codes. However, private companies do not exist as charitable institutions for my convenience. If they've got a good product, and people will pay money for it, they should obviously charge whatever the market will bear, and we should all buy their stock! If that bothers you, quit grumbling, and go build the better mousetrap . . . Finally, I am strongly in favor of the idea of publishing all calculational details required to reproduce the reported work (although that might require the reproducer to buy a proprietary program). However, I'd like to see this placed in supplementary material, as it is generally a total waste of journal space which remarkably few people will ever refer to. I speak with some authority on this, since in my earlier career as a synthetic organic chemist I contributed about fifty pages worth of garbage like "To a flame dried round bottom flask equipped with magnetic stirring was added blah blah blah" and I'm embarassed about it now. It's not that hard to get Suppl. Matl., or just write the author -- let's not make my shelves sag even more significantly than they already are. -- Christopher J. Cramer University of Minnesota Department of Chemistry 207 Pleasant St. SE Minneapolis, MN 55455-0431 (612) 624-0859 From balducci@serine.phr.utexas.edu Wed Jun 30 09:19:53 1993 Message-Id: <9306301920.AA10381@serine.phr.utexas.edu> To: CHEMISTRY@ccl.net Subject: Re: SAM1 reference and assorted comments following Date: Wed, 30 Jun 1993 14:19:53 -0500 From: Renzo Balducci I don't intend to join the discussion on proprietary vs. disclosed computational chemistry software. However, I just could not refrain from a comment on the following statement posted earlier by Dr. Graham Hurst. > I feel it is important to distinguish "science" from "code". The former > includes everything a researcher needs to know to reproduce numbers, eg. > equations, fundamental constants, fitted parameters. The latter implements > the former and will also have data structures, control structures, user > interface, programming techniques, performance tuning, etc. I completely disagree with this distinction. It is true that the actual "coding" (that is, the editing, compiling, testing, etc. of new software) is a necessary but collateral activity to the "scientific" research associated with the development of new computational methods. However, the definition of effective data structures and algorithms is, in my opinion, a crucial component of these research efforts. In fact, in several cases it would be impossible to even describe (much less to reproduce) the "science" of a method without a clear definition of the structure of the "coded" solution. Conventional mathematical idioms (such as equations and constants) are well suited for the description of some but not all problems of interest to the computational chemist. For instance, any non-trivial manipulation of the topology of molecular structures (ring perception comes to mind as an example) cannot be properly described without defining the representation model used. Such a description would in fact provide a specialized set of mathematical idioms necessary for the rigorous description of the algorithm and, therefore, instrumental in guaranteeing the reproducibility of the method. Conversely, any method representable with more conventional mathematical tools implicitly relies on a set of universally accepted data and code structuring techniques (e.g.: vector <--> 1D-array, matrix <--> 2D-array, summation <--> loop, equation <--> assignment-expression, parameters <--> function-arguments, etc.). The reproducibility of the results obtained with a published method is contingent upon the implicit or explicit definition of these "coding" techniques. I agree that in most cases a detailed description of the "code" would unnecessarily clutter the "science" of the paper. However, clever coding techniques aimed at the solution of previously unsolved problems or resulting in substantial performance boosts of existing solutions bear in my mind the same dignity and deserve the same recognition as any other original scientific achievement. In fact, in my opinion, the computational chemistry literature should devote a wider space to these type of findings, just like the literature of other computationally related sciences does. The advancement of science relies heavily of such a free interchange of ideas. Of course, I'm being naive here. If everybody did disclose their secret tricks how could anybody make a profit out of them :-) But, as I said, I don't want to get into this discussion... Regards, ---Renzo ----------------------------------------------------------------------- Renzo Balducci balducci@phenyl.phr.utexas.edu Research Associate & System Manager balducci@vax.phr.utexas.edu College of Pharmacy balducci@ccwf.cc.utexas.edu Univ. of Texas --- Austin, TX 78712 (512) 471-8509; FAX: 471-8762 ----------------------------------------------------------------------- From st-amant@cgl.ucsf.EDU Wed Jun 30 10:34:48 1993 Date: Wed, 30 Jun 93 17:34:48 -0700 Message-Id: <9307010034.AA24591@socrates.ucsf.EDU> From: st-amant@cgl.ucsf.edu (Alain St-Amant) To: chemistry@ccl.net, cramer@staff.tc.umn.edu Subject: Re: Just for argument's sake . . . Chris Cramer writes: > As for reproducibility, etc. Every scientific instrument in the > world can trace back to some researcher who built the first one. If I'm > one of those people, am I forbidden to publish until I allow a referee > to come use my machine or build his own?? Nonsense. If it's spectacular I think that the distinction some of us are making is that in one case, the author is WILLING to have people come use his machine while in the other, the person is UNWILLING to have people come use his machine because he is afraid of hurting his profits. And when we discuss this issue in computational chemistry, let's remember that our "machines" can be duplicated and distributed at virtually no cost. That's just one man's opinion, Alain St-Amant Department of Pharmaceutical Chemistry University of California, San Francisco From raman@bioc01.uthscsa.edu Wed Jun 30 16:50:31 1993 From: raman@bioc01.uthscsa.edu (C.S.RAMAN) Message-Id: <9307010250.AA01053@bioc01.uthscsa.edu> Subject: Re: Silicon Graphics To: nealn@candyman.esd.sgi.com (Neal Neuburger) Date: Wed, 30 Jun 1993 21:50:31 -0500 (CDT) > > >1-800-800-SGI1 or 1-800-800-7441 > > The above number is for SGI express. The main number is > > 800-800-4SGI or 800-800-4744 I am sorry, but, the latter number given by neal is devoted to Customer support. The 1-800-800-7441 does have an option for sales while the latter does not! I checked this out myself prior to posting. -- 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 UTHSCSA 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 d3f012@pellucidar.pnl.gov Wed Jun 30 11:54:42 1993 Date: Wed, 30 Jun 93 18:54:42 -0700 From: d3f012@pellucidar.pnl.gov Subject: MNDO, AM1, PM3, etc. To: chemistry@ccl.net Message-Id: <9307010154.AA08845@pellucidar.pnl.gov> I previously responded to the following paragraph from a posting by Andy Holder regarding the MNDO-family of methods: >4. It should be noted that, whatever is stated to the contrary, suff- > icient detail will be published on SAM1 so that other establishments > and individuals will be able to generate reproducible code. Note > that none of MJS Dewar's previous methods could have been coded > from scratch as many have claimed, (see above) without reference > to the code itself, either MOPAC or AMPAC. it is impossible to > obtain completely correct results for AM1 > and MNDO from the papers alone. Certain special corrections were > omitted from publication due to an oversight. (These corrections > form the subject of another paper to be released shortly.) Basically, I disagreed with his statement about not being able to reproduce the previous methods from the literature. I responded by delineating some of the typos I had found when I had implemented MNDO,AM1, and PM3 from scratch. The correction recently printed in JACS was indeed the typo in the nuc-nuc repulsion term for AM1 I had mentioned. So in defense of the original authors of the AM1 paper, it is indeed possible to correctly implement these methods from scratch without access to source code from AMPAC or MOPAC. One merely would have to see the obvious typo in the equation. Basically, everything in these original papers is very clearly described. Many thanks to Eve Zoebisch for FAXing the correction to me. I have another typo I discovered that I should report, in case others have accidentally implemented it. If this has already been printed as a correction, then please disregard my comment. The paper is Dewar and Thiel (TCA 46, (1977) p. 89-104) (a terrific paper by the way!). This is the paper where they give the details of calculating the MNDO two-electron integrals. eqn. 57 (page 103) for the [mu(z), mu(z)] interaction the last subexpression reads: [(R - D1a + a11^2]^-0.5 it should read: [(R - D1a + D1b)^2 + a11^2]^-0.5 Mark Thompson ************************************************************************** Mark A. Thompson Sr. Research Scientist email: d3f012@pnlg.pnl.gov Molecular Science Research Center FAX : 509-375-6631 Pacific Northwest Laboratory voice: 509-375-6734 PO Box 999, Mail Stop K1-90 Richland, WA. 99352 Argus available via anonymous ftp from pnlg.pnl.gov (130.20.64.11) (in the argus directory). Download the README file first. Disclaimer: The views expressed in this message are solely my own and do not represent Battelle Memorial Institute, Pacific Northwest Laboratory, or any of its clients. **************************************************************************