From Karyn_Kambur%3HARBOR.acad@autodesk Fri Aug 14 18:43:22 1992 Date: Fri, 14 Aug 92 14:34 PDT From: Karyn_Kambur@autodesk Subject: Welcome to HyperChem To: hyperchem@autodesk Welcome to the HyperChem users' e-mail group. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PURPOSE OF THE GROUP ==================== This e-mail group is designed primarily for, but not limited to, HyperChem users. Any and all scientific and technical issues related to the use of HyperChem are appropriate for discussion on this group, and we hope that it will become a valuable resource for HyperChem users. The group is unmoderated, so that any message you send to the group is sent automatically to all other members of the group. We trust that the group will remain unmoderated, so that the flow of ideas and information can be as uninterrupted as possible. In addition to messages, scripts, molecular mechanics parameters, and other additional material may be sent to the list. We do retain the right, however, to respond to the continued sending of irresponsible or offensive messages by removing the sender from the group. DISCLAIMER ========== The e-mail group is monitored by personnel from both Autodesk and Hypercube. Neither Autodesk nor Hypercube warrant, however, that they will answer all questions posted to the group. Autodesk and Hypercube personnel participate in the discussions in exactly the same manner as other group members. Any information on this group is provided "as is" without representation or warranty of any kind, either express or implied. The entire risk as to the use of this information is assumed by the user. ADDRESSES TO USE ================ To post a message to the group, please mail to hyperchem@autodesk.com. To be added to or removed from the group, please mail to hyperchem-request@autodesk.com We hope that you will find the discussion on the group to be informative and useful. Thank you for subscribing to the hyperchem e-mail group. Karyn Kambur (Autodesk) and Tom Slee (Hypercube). ________________________________________________________________________ From uucp@autodesk Fri Aug 14 21:48:05 1992 Date: Fri, 14 Aug 92 19:56:48 EDT From: SC18000 To: Subject: RE:HELP!? Dear Hyperchem netters, I have an old fashioned Zenith 386-16mHz machine. The one with the weird archecture and temdency toward keyboard hang-up. Because I came equipped with a pathetically small 40 Mbyte disk I'm using Stacker to attempt the storage of my too many projects. (They are all dearly productive to me and NO! I can't part with any single one of them.) My problem is that neither WINDOWS 3.1 nor Hyperchem runs properly with STACKER (SOB!). I can run some of the Windows stuff but some of it gives address and loading errors at run time. Hyperchem, with or without a properly installed hardware lock gives only run time loading errors. I don't even get to the message that tells me I should install the lock when I try to run it without the lock just to see. "A tale of woe and intrigue!" To quote Click and Clack (Tappit Bros.) Can anyone help in the solution to this existentially distructive problem? Sincerely (Desperately!) Ken Fountain (sc18@NEMOMUS) ________________________________________________________________________ From chapron%drac.CENG.cea.fr@autodesk Mon Aug 17 07:56:30 1992 Date: 17 Aug 92 12:41:13-2300 From: Yves chapron To: hyperchem@autodesk Subject: doc Please, could I have any general information on HyperChem ? Thank you Yves CHAPRON Biophysique moleculaire & cellulaire Grenoble C.E.N.G. 85 X 38041 GRENOBLE cedex FRANCE - Ph (33)76884212 Fx (33)76885487 E.M. chapron@drac.ceng.cea.fr ________________________________________________________________________ From mlynn%magnus.acs.ohio-state.edu@autodesk Mon Aug 17 11:57:55 1992 Date: Mon, 17 Aug 92 10:55:14 EDT From: Matthew A Lynn To: hyperchem@autodesk Subject: Metal-ring interactions Is there a way to draw and perform calculations on molecules like ferrocene where metal atoms interact with rings? I haven't found a way to make a dummy atom the centroid of a ring so I could "bond" the metal atom to it. I tried placing an iron atom between two cyclopentadienyl rings to see what would happen, but when I ran a geometry optimization, I got some erroneous atom-atom distances. Thanks for your help. Matt ________________________________________________________________________ From hurst Mon Aug 17 13:08:06 1992 To: hyperchem@autodesk.com, Matthew A Lynn Subject: Re: Metal-ring interactions > Date: Mon, 17 Aug 92 10:55:14 EDT > From: Matthew A Lynn > Subject: Metal-ring interactions > > Is there a way to draw and perform calculations on molecules like ferrocene > where metal atoms interact with rings? I haven't found a way to make a dummy > atom the centroid of a ring so I could "bond" the metal atom to it. I tried > placing an iron atom between two cyclopentadienyl rings to see what would > happen, but when I ran a geometry optimization, I got some erroneous > atom-atom distances. Thanks for your help. > Matt > This can be done for molecular mechanics if you define a new type and parameters. I would define a new type, "du" say, for a particular element (probably the non-element Lp?) in a copy of one of the type files (i.e. ambertyp.txt), add a type rule for it in chem.rul (note that bonds to Lp aren't counted in types so this shouldn't screw up the other type rules) add bond, angle and torsion parameters (preferably to a new parameter set created by copying and editing the standard files and editing chem.ini). You would then draw an Lp atom in the centre of the ring bonded to the 5 carbons and the iron. Graham ------------ Graham Hurst Hypercube Inc, 7-419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 internet: hurst@hyper.com ________________________________________________________________________ From kr%cup.portal.com@autodesk Wed Aug 19 08:02:18 1992 From: kr%cup.portal.com@autodesk To: hyperchem@autodesk Subject: periodic boundaries Date: Wed, 19 Aug 92 04:05:24 PDT First, I would like to say that it was a very good idea to establish this HyperChem email-group. Let me introduce myself shortly: The work I am involved in has to do with designing organic molecules that could serve as molecular building-blocks (MBB's) which can be polymerized into large lattice-like structures. Besides HyperChem, I have access to PolyGraf, both running on a Silicon Graphics machine, so I have the opportunity to compare these two products a little bit. My work of designing structures which can extend indefinitely into regular lattices in three dimensions naturally leads to the following question: Q.2 (8/19/92) Periodic Boundaries/Lattices In HyperChem (HC) it is possible to enclose an isolated molecule in a periodic boundary box, with or without surrounding water molecules as a solvent. Not so neat a feature is that the periodic boundaries can't be turned off from a menu but instead it has to be edited out from the .HIN text file manually (this is even mentioned in the user manual and I cannot find any other way on the menus). But more grave is that I have so far not been able to find out how to make the boundaries go through covalent bonds, which is important if one would like to simulate infinite polymers. The dialog box for boundary-setting seems to demand an isolated and nicely terminated molecular entity which can be entirely circumscribed by the boundaries and HC balks if one would like to make the boundaries smaller than the suggested minimum size. (For comparison: PolyGraf has a well developed periodic boundary system which can slice through anywhere. Also very convenient is the possibility of displaying all the 26 next-neighbour-cells to display the packing in e.g. a crystal. This array of cells is also editable. Very useful.) Have other users of HC stumbled on similar problems and how did they solve them ? I would appreciate any helpful comments. ------- End - of - Message ----------------------- (responsibility for my personal opinions is mine) Richard Stallman : "The more society becomes information instead of material objects, this question of whether a person can copy things will make the difference between a world of universal prosperity or a world of constant rat race." __________________________________________________ kr@cup.portal.com IIIII M M M M Markus Krummenacker I MM MM MM MM (working for I M M M M M M The Institute for I M M M M Molecular Manufacturing) I M M M M IIIII M M M M ________________________________________________________________________ From autodesk!hyper!slee Wed Aug 19 10:40:06 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Re: periodic boundaries To: hyperchem@autodesk Date: Wed, 19 Aug 92 10:09:26 EDT Markus Krummenacker writes: > In HyperChem (HC) it is possible to enclose an isolated molecule > in a periodic boundary box, with or without surrounding water > molecules as a solvent. > Not so neat a feature is that the periodic boundaries can't be > turned off from a menu but instead it has to be edited out from the > .HIN text file manually (this is even mentioned in the user manual > and I cannot find any other way on the menus). Actually, you can remove the periodic boundaries via the user interface, (though this is more of a workaround than a feature, and hence not documented). You bring up the periodic box dialog again, and click on the cancel button. Non-standard, of course, but it works. > But more grave is that I have so far not been able to find out how > to make the boundaries go through covalent bonds, which is important > if one would like to simulate infinite polymers. The dialog box for > boundary-setting seems to demand an isolated and nicely terminated molecular > entity which can be entirely circumscribed by the boundaries and HC > balks if one would like to make the boundaries smaller than the suggested > minimum size. > (For comparison: PolyGraf has a well developed periodic boundary system which > can slice through anywhere. Also very convenient is the possibility of > displaying all the 26 next-neighbour-cells to display the packing in e.g. > a crystal. This array of cells is also editable. Very useful.) > HyperChem's periodic box was designed for solution simulations. In the current version you cannot make the boundaries go through covalent bonds. Such a feature would be useful, of course, for both polymers and crystals. In the long run we certainly want to be able to handle extended systems, obviously, but there's nothing at the announcement stage about how long a run this is. > > Have other users of HC stumbled on similar problems and how did they solve > them ? > I would appreciate any helpful comments. Well, one out of two .... Tom Slee -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From hurst Wed Aug 19 14:21:29 1992 To: hyperchem@autodesk.com Subject: Re: periodic boundaries Tom Slee writes: > Markus Krummenacker writes: ... >> But more grave is that I have so far not been able to find out how >> to make the boundaries go through covalent bonds, which is important >> if one would like to simulate infinite polymers. The dialog box for >> boundary-setting seems to demand an isolated and nicely terminated molecular >> entity which can be entirely circumscribed by the boundaries and HC >> balks if one would like to make the boundaries smaller than the suggested >> minimum size. >> (For comparison: PolyGraf has a well developed periodic boundary system which >> can slice through anywhere. Also very convenient is the possibility of >> displaying all the 26 next-neighbour-cells to display the packing in e.g. >> a crystal. This array of cells is also editable. Very useful.) > > HyperChem's periodic box was designed for solution simulations. In the > current version you cannot make the boundaries go through covalent > bonds. Such a feature would be useful, of course, for both polymers > and crystals. In the long run we certainly want to be able to handle > extended systems, obviously, but there's nothing at the announcement > stage about how long a run this is. HyperChem can simulate infinite systems (within the nearest image approximation) with molecular mechanics, but the display is not ideal for bonds that cross the boundary. The dialog box in HyperChem does insist that all atoms lie within the unit cell before it creates a periodic box. This is done to insure that atoms are not multiply specified since the first step of periodic calculation is to translate all atoms into the unit cell, so that nearest images can be simply calculated. If the atoms in the unit cell were displayed, bonds that cross the boundary would be shown as long bonds going through the box (i.e. not between nearest images of atoms, but between the images within the unit cell). This display problem is somewhat overcome in HyperChem by translating each terminal atom (like hydrogens) to the image nearest to the atom that it is attached to. For non-terminal atoms long bonds are displayed. Thus a bond that "crosses" a boundary could be drawn to the image inside the box, rather than the one outside the box that "looks right", to set up the simulation. For display, atoms (and bonds) can be be drawn outside the box after it is created. If you then do a mechanics optimization with zero cycles (equivalent to a single point but updates display) the non-terminal atoms will be translated into the box and shown with long bonds. I hope that helps... Graham ------------ Graham Hurst Hypercube Inc, 7-419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 internet: hurst@hyper.com ________________________________________________________________________ From mlynn%magnus.acs.ohio-state.edu@autodesk Thu Aug 20 14:13:17 1992 Date: Thu, 20 Aug 92 13:33:42 EDT From: Matthew A Lynn To: hyperchem@autodesk Subject: Ball and stick rendering In the Hyperchem brochure there is a drawing of crown ether in ball and stick rendering. In our software package I find no mention of this type of rendering. Is ball and stick just a mixing of spheres and sticks? How can I get drawings in balls and sticks? Thanks for your help. Matt ________________________________________________________________________ From autodesk!hyper!slee Thu Aug 20 16:11:38 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Re: Ball and stick rendering To: hyperchem@autodesk Date: Thu, 20 Aug 92 15:15:32 EDT > > In the Hyperchem brochure there is a drawing of crown ether in ball and stick > rendering. In our software package I find no mention of this type of > rendering. Is ball and stick just a mixing of spheres and sticks? How can I > get drawings in balls and sticks? Thanks for your help. > Matt > The ball and stick rendering is available on the SGI version of HyperChem, but not on the Windows version. In the SGI version we were able to take advantage of their GL graphics library to do all kinds of things that are much less practical under Windows. Tom Slee -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From pbays%saintmarys.edu@autodesk Fri Aug 21 18:52:32 1992 Date: Fri, 21 Aug 1992 16:20:52 +0500 () From: Phil Bays Subject: Torsion Driver To: hyperchem@autodesk I have just begun to use Hyperchem. I have not found any reference to a torsion driver. Is it possible to do a conformational search about one or more dihedral angles? Perhaps I have not read far enough in the manuals?? Good documentation, by the way! Phil Bays ________________________________________________________________________ From pbays%saintmarys.edu@autodesk Mon Aug 24 18:03:14 1992 Date: Mon, 24 Aug 1992 13:08:42 +0500 () From: Phil Bays Subject: Torsion Search To: hyperchem@autodesk I have just begun to use Hyperchem. I have not found any reference to a torsion driver. Is it possible to do a conformational search about one or more dihedral angles? Perhaps I have not read far enough in the manuals?? Good documentation, by the way! Phil Bays ________________________________________________________________________ From batt%kenyon.edu@autodesk Mon Aug 24 18:03:15 1992 Date: Mon, 24 Aug 1992 15:39:34 EDT From: "R. BATT" To: HYPERCHEM@autodesk Subject: ball and stick Your answer to recent inquiry about ball and stick models made argument that such would be impractical (too slow) on PC. I think the issue of being too slow should be left to user, who may well be willing to wait for a ball and stick model to be displayed. In other words, you should provide the option, and let the user decide if she/he wants to use it. Alchemy III does ball and stick for small molecules in reasonable times on 25 and 33 MH machines. ________________________________________________________________________ From autodesk!hyper!slee Mon Aug 24 21:11:49 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Re: Torsion Driver To: hyperchem@autodesk Date: Mon, 24 Aug 92 9:48:56 EDT Phil Bays writes: > I have just begun to use Hyperchem. I have not found any reference to a > torsion driver. Is it possible to do a conformational search about one or > more dihedral angles? Perhaps I have not read far enough in the manuals?? There is no torsion driver currently included in HyperChem, but despair not! If you have a copy of the spreadsheet program Excel 4.0, and are working with HyperChem for Windows, then there is a macro available (free) that drives two dihedral angles and produces graphs of the results. It offers three computational options for molecular mechanics: i) single point calculations at each value, ii) geometry optimization at each value, with torsions restrained to remain close to specified values, iii) as (ii), but with the angle reset to precisely the specified value, for a final single point calculation. Of these, only (i) is available for the quantum mechanics methods at present. If you have access to CompuServe, this macro is posted there in a zipfile (Autodesk software forum, section 8). If you don't, just e-mail a request (to hyperchem-request@autodesk.com) and I will send you a uuencoded version by e-mail. Please let me know whether you have the pkunzip utility when you send the request. Tom Slee -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From autodesk!hyper!slee Tue Aug 25 10:44:24 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Re: ball and stick To: hyperchem@autodesk (HyperChem e-mail group) Date: Tue, 25 Aug 92 9:42:14 EDT Russell Batt writes: > > Your answer to recent inquiry about ball and stick models made argument that > such would be impractical (too slow) on PC. I think the issue of being too slow > should be left to user, who may well be willing to wait for a ball and stick > model to be displayed. In other words, you should provide the option, and let > the user decide if she/he wants to use it. Alchemy III does ball and stick > for small molecules in reasonable times on 25 and 33 MH machines. Point taken! We are currently assessing what features we can put into the next version (while working on others) and this input is valuable. Tom -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From autodesk!hyper!slee Wed Aug 26 19:44:57 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Torsion driving on the SGI To: hyperchem@autodesk (HyperChem e-mail group) Date: Wed, 26 Aug 92 14:08:47 EDT Following the recent question concerning torsion driving in HyperChem, I have just written a C Shell script for HyperChem on the SGI that carries out torsion driving on a selected torsion angle, using single-point or restrained geometry optimization calculations at each step. It's pretty simple and contains instructions for use. Elaborations to more angles would be fairly straightforward. Anyone interested please send a message to hyperchem-request@autodesk.com Tom Slee -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From pbays%saintmarys.edu@autodesk Thu Aug 27 17:43:05 1992 Date: Thu, 27 Aug 1992 15:37:10 +0500 () From: Phil Bays Subject: System crash To: hyperchem I am running hyperchem on an indigo using 4.0.5MM (for the video board). I have 48 Mb memory. NOw I was doing a relatively "simple" geomentry optimization on phenol in a box of water. The calculation had run for about 20 hours. I was simulataneously playing music on the CD and running other applications. At one point I called the HyperChem icon to see where the calculation stood and the system crashed with the following error message: Irix killed due to memory error between phys. addresses 0x1C00000 and 0X1FFFFFF, PC: 0x800045E4 ep: 0xFFFFEC80. When I brought the system back up I was told that: savecore /usr/adm/crash/umcore0. I looked in /usr/adm/crash/ and usre enough a huge thing exists called umcore0 as well as unix0. Now the question: Why did this happen, and can I trash the two files in crash? I might also add that since installing HyperChem I have had several problems of this sort. Twice when I tried to log off, the screen went blank blue, leaving a black box when the tool box should be, and then hung. Any advice or comments? Phil Bays ________________________________________________________________________ From Bill_Glauser@autodesk Thu Sep 3 22:09:36 1992 From: Bill_Glauser@autodesk Date: Thu, 3 Sep 92 18:27 PDT Subject: entropy calculations using MD To: hyperchem@autodesk I am a long-time listener, first-time caller. Molecular modeling packages, such as HyperChem, offer the possibility of computing thermodynamic quantities from molecular dynamics simulations using the machinery of statistical mechanics. By generating a series of configurations consistent with a Boltzmann distribution, one may sample any number of quantities at each time step. For mechanical quantities, such as internal energy or viscosity, sampling is straightforward. For non-mechanical quantities, such as the entropy, my intuition does not fare quite so well. If one could calculate the partition function from an MD (or MC) simulation then S = U/T + k ln Q and the problem would be solved. I wonder if one can reasonable employ a direct counting of states (density of states at constant energy) and use the statistical definition of entropy: S = k ln W This would also be great for teaching purposes, even though it may only provide a crude estimate of S. I would appreciate any and all comments and insight. _______________________________________________________________ Bill Glauser Scientific Modeling Division Autodesk, Inc. billgl@autodesk.com ________________________________________________________________________ From DSMITH%uoft02.utoledo.edu@autodesk Mon Sep 7 14:55:11 1992 Date: 07 Sep 1992 13:49:10 -0400 (EDT) From: "DR. DOUGLAS A. SMITH, UNIVERSITY OF TOLEDO" Subject: cyclic structures for teaching/research To: hyperchem@autodesk We are in the planning stage of preparing a library of cyclic structures for use with Hyperchem. Initially, we are planning to use those structures which appear in a standard organic textbook, Organic Chemistry by T. W. G. Solomons, 5th edition. However, we would like to solicit help and advice from Hyperchem users. There are many structures in this textbook, and in order for us to do a timely and useful job, we must choose some limited subset of cyclic structures, to which we can add later. What structures would users think are most important, and which are least important? Is it more useful to have polycyclic aromatic hydrocarbons done first, or biological molecules? Drugs or architecturally interesting (e.g. cubane, dodecahedrane) molecules? Natural products? Please send your opinions directly to me and I will summarize for the net. Thank you in advance. 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 reid%cvax.psi.ch@autodesk Wed Sep 16 09:52:59 1992 From: reid%cvax.psi.ch@autodesk Date: Wed, 16 Sep 1992 14:53:18 +0200 To: hyperchem Subject: Hyperchem wishlist Is there something wrong with my mail path in this list? I've only received two messages so far. Having used Hyperchem for a couple of months now, I have a couple of questions. Actually, since all my experience and searches of the manuals suggest that the answers are mostly "No", this is really a wish-list to those nice developers at Hypercube. a) Is there any way of typing in the coordinates for an atom (a la Autocad) rather than just clicking the mouse? Actually, I far prefer most of the way Autocad does things (mirroring, copying, insertion, etc). It was a real pain to make a C70 molecule out of the supplied C60.HIN. In Autocad you just orient the C60 correctly, lop off 20 atoms and their bonds, and reflect everything about the equator. None of the Hyperchem editing actions behave in the same way as Autocad. e.g. you can't retain the original objects after a mirroring operation, you can't insert to a specific location, it's very difficult (impossible?) to move a selection set to a specific location. And so on. b) Is it possible to read in data from a file other than a HIN file? e.g. a MOPAC/AMPAC-style file or just a set of atom types and coordinates (again, a la MOPAC). c) Does Hyperchem use symmetry relations? I suspect it might since a calc- ulation on corannulene went far faster than for the radical with a H atom added to one of the central carbons. If so, how good is it at recognising symmetry (e.g. all the symmetries of the C70 molecule). d) Is it possible to get hyperfine coupling parameters? I can't find any reference to hfc's in the manuals at all. I'm mainly interested in radicals (specifically muonated, but protium is a good approximation). Enough for now. I hope none of these are FAQs, but since I've only seen two questions since 4th September, I don't think they are. ivan ________________________________________________________________________ From hurst Wed Sep 16 15:26:00 1992 To: hyperchem@autodesk.com Subject: Re: Hyperchem wishlist Cc: reid@cvax.psi.ch reid@cvax.psi.ch writes: > Is there something wrong with my mail path in this list? I've only > received two messages so far. If you get this twice (once directly and once via hyperchem@autodesk.com) everything should be okay. Traffic has been a little quiet lately... > > Having used Hyperchem for a couple of months now, I have a couple of > questions. Actually, since all my experience and searches of the manuals > suggest that the answers are mostly "No", this is really a wish-list to those > nice developers at Hypercube. > a) Is there any way of typing in the coordinates for an atom (a la Autocad) > rather than just clicking the mouse? Actually, I far prefer most of the way Yes. Select the atom and use Edit/Translate or double click on either of the translation tool icons. If you want to independently move atoms that are bonded make sure that "Whole molecule translation" is off in the Tools Preferences dialog (or chem.ini). > Autocad does things (mirroring, copying, insertion, etc). It was a real pain Ask your dealer if (s)he can get a copy of MoleCAD. John Walker wrote it so he could draw molecules in AutoCAD and move them to HyperChem, but I don't know its current status. > to make a C70 molecule out of the supplied C60.HIN. In Autocad you just > orient the C60 correctly, lop off 20 atoms and their bonds, and reflect > everything about the equator. None of the Hyperchem editing actions behave > in the same way as Autocad. e.g. you can't retain the original objects after > a mirroring operation, you can't insert to a specific location, it's very > difficult (impossible?) to move a selection set to a specific location. And > so on. You can do exactly what you want (i.e. delete, select, copy, paste, translate, and relflect will accomplish it) but the philosophy of HyperChem is different from AutoCAD - since molecular structures tend to follow rules, let the program (HyperChem) create the 3D structure! If you sketch the 2D topology of C70 (start with a central 5 membered aromatic ring, surround with 6 membered aromatic rings, etc.) and then model build you will get a *better* C70 structure than by the process you describe! Each of the issues you raise *can* be done in HyperChem, though you are more familiar with the AutoCAD way. Reflect doesn't leave a copy of the original, but you can get a copy by using Copy and Paste beforehand (Paste will offset the "copy" but you can translate it back by using the original to define POINT and then translate the selection back to POINT). You *can* translate a selection to a specific location - that's precisely what Translate Selection in the Translate dialog box does! > b) Is it possible to read in data from a file other than a HIN file? e.g. > a MOPAC/AMPAC-style file or just a set of atom types and coordinates (again, > a la MOPAC). In this version you are limited to HIN and PDB, but these are both ASCII and are essentially atom types and Cartesian coordinates. More formats are planned for future releases, but you can get 3rd party file converters now (for instance ConSystant from ExoGrpahics, P.O. Box 655, West Milford, NJ, USA 07480). > c) Does Hyperchem use symmetry relations? I suspect it might since a calc- > ulation on corannulene went far faster than for the radical with a H atom > added to one of the central carbons. If so, how good is it at recognising > symmetry (e.g. all the symmetries of the C70 molecule). No it doesn't. The time difference is more likely closed shell vs. open shell or RHF vs. UHF. > d) Is it possible to get hyperfine coupling parameters? I can't find any > reference to hfc's in the manuals at all. I'm mainly interested in radicals > (specifically muonated, but protium is a good approximation). Almost. If you do a UHF calc'n on a radical, the logfile will have alpha and beta spins for each atom. Subtract these and multiply by a per element value from the literature (I think they are tabulated in Pople & Beveridge's book "Approximate Molecular Orbital Theory"). > Enough for now. I hope none of these are FAQs, but since I've only > seen two questions since 4th September, I don't think they are. > > ivan > Only one "no"! Hope this helps! Graham ------------ Graham Hurst Hypercube Inc, 7-419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 internet: hurst@hyper.com ________________________________________________________________________ From Bill_Glauser@autodesk Wed Sep 16 22:23:34 1992 From: Bill_Glauser@autodesk Date: Wed, 16 Sep 92 18:43 PDT Subject: benchtop vs theoretical chemists To: hyperchem@autodesk I had a random philosophical thought regarding expectations about what may be reasonably accomplished with molecular modeling techniques. Conventional wisdom (or at least my first impression) states that theoretical/computational chemists would be the most difficult and discerning audience with respect to purveyors of fine software. Experimentalists, in contrast, should have a less discerning palette in this regard. Yet, the opposite is probably true. Theoreticians appreciate the inherent limitations of computational algorithms as currently implemented, and has thus circumscribed their expectations. Experimentalists, on the other hand, want answers to real world problems that are the most difficult to obtain. For example, a theory person generally makes due with gas phase results for minimum- energy or transitional structures, yet experimentalists may insist upon the solvated analog. Further, the quantities most desired by experimentalists, such as hydrophobicity, solubility, reactivity, are precisely those that are hardest to computationally model. Where modeling works the best is in the area of aiding and guiding the intuition. These methods offer the possibility of giving qualitatively and sometimes semi-quantitative answers to carefully thought out questions. Sometimes trends are more important than the accuracy of a given point. Let me know if there is any merit to my hypothesis. I would appreciate any comments especially from more experimentally minded folks. ------------------------------- Bill Glauser Scientific Modelling Division Autodesk, Inc. billgl@autodesk.com -------------------------------- ________________________________________________________________________ From CHUCK%psipsy.uct.ac.za@autodesk Thu Sep 17 06:11:35 1992 To: hyperchem@autodesk From: "Charles Marais" Date: 17 Sep 92 11:40:01 SAST-2 Subject: Huckel parameters Hi I've had the PC version of Hyperchem for a couple of days now, and it looks excellent ( but, of course, I'll post my wish list soon). More immediately, we'd like to look at some organometallic complexes with EH calcs, and have double zeta params from Hoffman's group available for most of the periodic table. However, looking at the (very small !?) Hyperchem parameter file, it seems that Hyperchem uses single zeta only. Is there a way of properly using our parameters with Hyperchem ? Charles Marais Department of Chemistry UCT Private Bag Rondebosch South Africa 7700 chuck@psipsy.uct.ac.za ________________________________________________________________________ From autodesk!hyper!slee Thu Sep 17 11:53:31 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Re: Huckel parameters To: hyperchem@autodesk (HyperChem e-mail group) Date: Thu, 17 Sep 92 9:34:46 EDT > More immediately, we'd like to look at some organometallic complexes > with EH calcs, and have double zeta params from Hoffman's > group available for most of the periodic table. However, looking at > the (very small !?) Hyperchem parameter file, it seems that Hyperchem > uses single zeta only. > Unfortunately the answer is no in the current version. HyperChem version 2.0 is limited in its semi-empirical methods to atoms up to Argon. In the next release we will have the integrals required for at least the first row of transition metals, and will be using double zeta functions in the "standard" way. (what Hoffmann does being the definition of standard in EHMO calculations). We will also have quite a lot of other new semi-empirical functionality. Tom Slee -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From DSMITH%uoft02.utoledo.edu@autodesk Sun Sep 20 23:40:48 1992 Date: 20 Sep 1992 22:49:58 -0400 (EDT) From: "DR. DOUGLAS A. SMITH, UNIVERSITY OF TOLEDO" Subject: sci.chem newsgroup debate To: hyperchem@autodesk I just posted this reply to a rather one sided and nasty posting in the sci.chem newsgroup, and thought I might share it with all of you. Doug Smith ============================================================================ X-News: uoft02 sci.chem:7125 From: dsmith@uoft02.utoledo.edu Subject:Re: 3D modeling software Date: 20 Sep 92 22:46:38 EST Message-ID:<1992Sep20.224638.10044@uoft02.utoledo.edu> In article <1992Sep18.223529.4567@syma.sussex.ac.uk>, kafc6@syma.sussex.ac.uk (Kenneth Seddon) writes: > I must say that I do not feel that sci.chem is > the correct forum for blatant advertising! Probably correct, but this is an unmoderated list and opinions that are useful, informative, and avoid ad hominum attacks are worth posting, regardless of commercial content. > However, if you insist in pushing a commercial > product in an essentially academic environment, That is interesting. I have never done or seen a statistical breakdown of posts as to source, i.e. academic vs. industrial vs. private. Has anyone else? If so, please post it. Without that, I would say that there is certainly a significant amount of non-academic traffic in this group. > you must live with criticisms of it. Very true. Email and newsgroups are not in the least private, and just like memos, can get you into real trouble. > Having > recently been asked to evaluate Hyperchem for > a retail organization, I found it slow, > inefficient, and clumsy, with misleading advertising > (verging on complete misrepresentation). Very interesting. I was a beta tester for Hyperchem, and my entire research group found it easy to learn and use, fast, effective, etc. Even though we do most of our research on Crays and big workstations, we have Hyperchem. I wonder what kind of hardware you used. Right now, I use a 485/50MHz with 32 MB of RAM and a 1.4 GB hard disk, all from Dell Computers. However, we did the beta testing on a 33 MHz 386 with an 80387 coprocessor and only 8 MB of RAM. I even use the program on my 386 N+ Dell notebook, with 80387 coprocessor and 4 MB of RAM and find it very useable. Maybe you were just underpowered. One does have to remember to do the right types of chemistry on the right program on the right hardware, and if you want to do protein dynamics (even though this is what Hyperchem suggests you can do) I would not use Hyperchem. > Readers > of this newgroup would be advised to give it a wide > berth, and to pay attention to the considered comments > of Henry Rzepa, rather than vested commercial interests. Actually, readers of this newsgroup and any other scientists, potential molecular modellers, etc., would be advised to get a trial copy of this or any other program, try it out on their own hardware and research problems, and make their own informed decision. My decision is that Hyperchem is a useful and in some cases valuable tool for research, even when I have many other resources available. 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 autodesk!hyper!burton Mon Sep 21 12:25:43 1992 From: autodesk!hyper!burton (Ian Burton) Subject: Re: DEC version? To: hyperchem@autodesk Date: Mon, 21 Sep 92 11:50:54 EDT Kevin, In reply to your message: > >Does hyperchem run on the DEC 5000 workstations? We have a user who is >in need of a supported package which runs on a 5000. Any info would be >appreciated. > >++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >|| Kevin Moore North Carolina Supercomputing Center || >|| Scientific Support Analyst 3021 Cornwallis Rd. || >|| (919) 248-1179 Research Triangle Park, NC 27709 || >++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > At present the only supported platforms are MS-DOS compatible PC's running Windows 3.0 and higher, and SGI Workstations. Other platforms are being considered for future releases, but for the present the DEC 5000 workstation is not supported. Ian Burton. -- Ian Burton Hypercube, Inc. Internet: burton@hyper.com #7-419 Phillip St. Telephone: (519) 725-4040 Waterloo, Ontario Facsimile: (519) 725-5193 Canada N2L 3X2 ________________________________________________________________________ From CHUCK%psipsy.uct.ac.za@autodesk Tue Sep 22 11:24:06 1992 To: hyperchem@autodesk From: "Charles Marais" Date: 22 Sep 92 15:14:24 SAST-2 Subject: Spin densities I have a question about spin densities - I'm not sure what they mean. I'm looking at methoxybenzene radical cation derivatives, and am using AM1/UHF to look for sites for radical attack. Now, the manual says the two-color spin density contour tells you the probability of finding more 'up' than 'down' spin at a certain point, and the countours also show negative values, which, although a probability can't be <0, I take to mean the other spin dominates. But I'm looking for a certain('unpaired' or 'excess' ) spin in the radical. Which color is it ? Another thing is, these regions are very distinctly demarcated in certain planes, and DON'T ALWAYS LOOK LIKE THE ALPHA HOMO in that plane, which I expected would be where the unpaired spin lives. What's up ? Any help with interpreting this would be much appreciated. C Charles Marais Department of Chemistry UCT Private Bag Rondebosch South Africa 7700 chuck@psipsy.uct.ac.za ________________________________________________________________________ From autodesk!hyper!slee Tue Sep 22 13:43:53 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Re: Spin densities To: hyperchem@autodesk Date: Tue, 22 Sep 92 12:54:55 EDT Following up your spin-density questions: > > I'm looking at methoxybenzene radical cation derivatives, and am > using AM1/UHF to look for sites for radical attack. > Now, the manual says the two-color spin density contour tells you the > probability of finding more 'up' than 'down' spin at a certain point, > and the contours also show negative values, which, although a > probability can't be <0, I take to mean the other spin dominates. But > I'm looking for a certain('unpaired' or 'excess' ) spin in the > radical. Which color is it ? You are right that we do mean that the other spin dominates. The positive (alpha spin, or majority spin) density is green. > > Another thing is, these regions are very distinctly demarcated in > certain planes, and DON'T ALWAYS LOOK LIKE THE ALPHA HOMO in that > plane, which I expected would be where the unpaired spin lives. > What's up ? > > Any help with interpreting this would be much appreciated. > > Charles Marais > Department of Chemistry > UCT Private Bag > Rondebosch > South Africa 7700 chuck@psipsy.uct.ac.za > I did a few calculations, together with Yufei Guo here at Hypercube, and have a few comments on your spin density observations that may help. First, we did an AM1/RHF calculation on methoxy benzene cation radical, where the spin density should look very much like the HOMO. We found that it did, with most spin density at carbon 4, some at carbon 2 and some at carbon 1 of the benzene ring. Here the spin density is all positive (green), as there is no spin polarization possible within the RHF model. The HOMO is very similar to the spin density, as you would expect. The slight differences are because HyperChem uses the NDO approximation in the calculation of molecular properties from the wavefunction, including the spin density, but not in calculating the molecular orbital densities. (for details, see page 82 to 84 of the Theory and Methods section of the Computational Chemistry Handbook). (As an aside, you should ensure that your plane offset is not zero when plotting these graphs, as both the RHF spin density and the HOMO have a node in that plane, and all you get is noise. A plane offset of 0.5 or 1 A gives perfectly okay results.) Next we did your UHF calculation. Here the spin density is slightly different from the RHF case: the spin polarization produces regions of negative (violet) spin density at carbons 3 and 5, which makes sense. These regions are smaller in magnitude than the positive regions, which are still centred at carbons 1,2, and 4. Finally, we looked at the UHF MO's. You are right that the alpha-electron HOMO does not look like the spin density plot--but the strict analogy is broken by the UHF treatment, as the spin density now has contributions from all molecular orbitals. I don't have enough experience to know how much contribution you would expect from each orbital, but in the quasi-degenerate pi system of benzene derivatives (and the spin density is primarily pi) it seems not unreasonable that you would get significant contributions from several of the orbitals, their similar energies leading to significant mixing under the influence of the spin polarization. I'm no expert at radical chemistry, but my suggestion would be to take the RHF result as your "connection to orbital-picture intuition", as this is where the HOMO-spin density analogy comes from, look at the UHF spin density to see how this is modified by spin-polarization, and ignore the UHF MO's as too mixed up to yield simple interpretations. Tom Slee -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From CHUCK%psipsy.uct.ac.za@autodesk Thu Sep 24 08:42:26 1992 To: hyperchem@autodesk From: "Charles Marais" Date: 24 Sep 92 13:45:55 SAST-2 Subject: Some talk about Hyperchem SOME TALK ABOUT HYPERCHEM A couple of years ago, I had some very disappointing experiences with stuff you could buy - I tested four or five packages - they were just buggy and broken, and the companies lied about what it could do. A particularly interesting occurance at that time was the entry of a company into this market with THE BEST graphics and fastest computation I've ever seen on a PC, running on a Mac/Opus coprocessor combination. They wouldn't even disclose the computational methods used in the system ! Needless to say, we were (again) 'unable to order'. Hyperchem seems to break this 'closed' mindset, I would say. After using Hyperchem for a week or so on a 25 Mhz 486 for research work together with MOPAC, I think it's certainly the best I've seen on PC's. This goes for documentation as well. Also, the authors were helpful (for now - I guess they'll not be able to handle as much traffic from everyone as they got from me eventually !) and informed - something which is extremely important for commercial software in an academic environment - academics want to see commercial work validated, somehow. So far, no modeling package can do what users want all of the time. This has always bothered me, because it is possible. Hyperchem is the first (I think !) modeling package to offer data and control (but still not enough) to another running program. I believe that the completeness of DDE-like implementations in modeling packages will determine which one wins. Some things are perplexingly missing from Hyperchem (see 3) and 10) ), and I reckon these should immediately be fixed, before a next release. Here's my wishlist. 1. Selection of the default element during building requires too many steps. Why not pop up a small periodic table so that this can be done on the fly ? 2. I have a i860 Numbersmasher board from Microway. What are the chances of getting back-end computation on this board ? Microway have vectorizers and all... This would be GREAT. It would be nice if some discussion with Hypercube could be initiated about most popular back ends for PC users. 3. A Serious Omission is the lack of the capability wrt reading/writing other file formats a) Macromodel b) Mopac (especially) 4. Simple annotation before printing - bond angles, lengths, torsions etc 5. Aborting an EP or MO plot calc should not thereafter require recalc of whole wavefunction. 6. Option to save MO's so that recalc is not neccesary later. 7. Option to set a damping factor in minimizations. 8. 3D orbital plots and densities on PC's. What if users want the pictures, no matter how long it takes ? Hyperchem authors perhaps underestimated the informative content and general coolness of these. 9. Density subtractions for difference maps wrt two different computations on the same molecule. 10. A Crippling Omission. Where did the "only up to Cl" limitation on semi-empirical calculation come from ? The whole periodic table is available for EH calcs, and most of the p-block for PM3 and AM1. Adding these elements should do nothing to the speed or to the memory requirements, as far as I can see (all the methods are valence shell only). I would be very interested to learn why exactly this decision was taken. It just seems very strange. 11. Please don't force users to rename a parameter file for any of the methods. Let them fool around freely with parameter sets, in files of their choice, selectable at run-time. 12. Allow modification of force-field and semi-empirical parameters on the fly in a dialog box. Allow these to be saved. 13. A Pretty Serious Omission. Offer, via DDE, data somewhat closer to the computing methods : a) gradient components b) the capability to draw simple graphics, like arrows, lines etc c) orbital coefficients This would probably spawn the most successfull and prolific user group in this spiral arm of the galaxy. I rank my most important 5 points above : 10, 3, 13, 12, 2. And that's it from me, you'll be thankfull to know. Charles. Charles Marais Department of Chemistry UCT Private Bag Rondebosch South Africa 7700 chuck@psipsy.uct.ac.za ________________________________________________________________________ From hurst Thu Sep 24 15:42:59 1992 To: "Charles Marais" Subject: Re: Some talk about Hyperchem Cc: hyperchem@autodesk Thanks for the wishlist items! Most of the things are already on our list, but with finite resources we had to draw the line somewhere or you'd still be waiting for good modelling on the PC! I can respond to some of your points: > 1. Selection of the default element during building requires too many steps. > Why not pop up a small periodic table so that this can be done on the fly ? This is included in the SGI version, and a Visual Basic application that does this for the windows version has been written by Tony Brand. I think it is in the HyperChem library of Autodesk's compuserve forum, and you can get a zipped version by email by asking hyperchem-request@autodesk.com (you might want to ask for the Excel 4.0 macro that drives two dihedral angles and produces a conformational surface plot). > 2. I have a i860 Numbersmasher board from Microway. What are the chances of > getting back-end computation on this board ? Microway have vectorizers and > all... This would be GREAT. It would be nice if some discussion with > Hypercube could be initiated about most popular back ends for PC users. This is of interest, at least to me. Previous (pre-Autodesk) versions of HyperChem supported arrays of transputers in PCs. > 3. A Serious Omission is the lack of the capability wrt reading/writing other > file formats a) Macromodel b) Mopac (especially) More formats will be supported by HyperChem in the next release. Right now though you can buy a file (and clipboard) converter from ExoGraphics, P.O. Box 655, West Milford, NJ 07480, USA (Phone 201-728-0188, FAX 201-728-0735) for $199. > 4. Simple annotation before printing - bond angles, lengths, torsions etc Nice idea but not trivial. We recommend annoting an image (bitmap or drawing) in a seperate Windows app. To get a complete list of internal coordinates, do an MM+ single point with a log file and PrintLevel=1 in chem.ini (or after running script command "mechanics-print-level 1"). > 6. Option to save MO's so that recalc is not neccesary later. This is addressed to some extent in the SGI version. The grid of points used to generate the orbital isosurface can be saved to a file. > 7. Option to set a damping factor in minimizations. I don't know what you mean by this? Are you talking about geometry optimization or the damping of SCF iterations as a convergence enhancing technique? > 8. 3D orbital plots and densities on PC's. What if users want the pictures, > no matter how long it takes ? Hyperchem authors perhaps underestimated the > informative content and general coolness of these. We'd love this too, but it's a lot of software coding for us that is done by the OS and hardware on the SGI. Some customers are exploring using Autodesk's 3D Studio for this (but I'm embarrased that I've forgotten who!)... > 10. A Crippling Omission. Where did the "only up to Cl" limitation on > semi-empirical calculation come from ? The whole periodic table is available > for EH calcs, and most of the p-block for PM3 and AM1. Adding these elements > should do nothing to the speed or to the memory requirements, as far as I can > see (all the methods are valence shell only). I would be very interested to > learn why exactly this decision was taken. It just seems very strange. Unlike ab initio, where linear combinations of "1s" gaussians are used to created the 2s, 3s, etc., using Slaters in semi-empirical calculations requires writing more code for integrals (and their derivatives) for each increase in principal quantum number. Since our code is all new (in C) this requires significant effort in development and testing, so that's why there is the limitation. It's coming though... > 11. Please don't force users to rename a parameter file for any of the > methods. Let them fool around freely with parameter sets, in files of their > choice, selectable at run-time. You must be referring to quantum mechanics, because we have made it very easy to add or change molecular mechanics parameters. I guess we didn't think people played with semi-empirical parameters as much... (PM3 was a late addition.) > 13. A Pretty Serious Omission. > Offer, via DDE, data somewhat closer to the computing methods : > a) gradient components > b) the capability to draw simple graphics, like arrows, lines etc > c) orbital coefficients > This would probably spawn the most successfull and prolific user group in > this spiral arm of the galaxy. I agree with a) and c), but for b) find a graphics progam that talks DDE and has a macro language. It could tell HyperChem to copy the image, paste it and then edit it. Thanks for the feedback! Graham ------------ Graham Hurst Hypercube Inc, 7-419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 internet: hurst@hyper.com ________________________________________________________________________ From autodesk!hyper!hurst Thu Sep 24 17:45:34 1992 Date: Thu, 24 Sep 92 15:43:00 -0400 From: autodesk!hyper!hurst (Graham Hurst) To: "Charles Marais" Subject: Re: Some talk about Hyperchem Cc: hyperchem@autodesk Thanks for the wishlist items! Most of the things are already on our list, but with finite resources we had to draw the line somewhere or you'd still be waiting for good modelling on the PC! I can respond to some of your points: > 1. Selection of the default element during building requires too many steps. > Why not pop up a small periodic table so that this can be done on the fly ? This is included in the SGI version, and a Visual Basic application that does this for the windows version has been written by Tony Brand. I think it is in the HyperChem library of Autodesk's compuserve forum, and you can get a zipped version by email by asking hyperchem-request@autodesk.com (you might want to ask for the Excel 4.0 macro that drives two dihedral angles and produces a conformational surface plot). > 2. I have a i860 Numbersmasher board from Microway. What are the chances of > getting back-end computation on this board ? Microway have vectorizers and > all... This would be GREAT. It would be nice if some discussion with > Hypercube could be initiated about most popular back ends for PC users. This is of interest, at least to me. Previous (pre-Autodesk) versions of HyperChem supported arrays of transputers in PCs. > 3. A Serious Omission is the lack of the capability wrt reading/writing other > file formats a) Macromodel b) Mopac (especially) More formats will be supported by HyperChem in the next release. Right now though you can buy a file (and clipboard) converter from ExoGraphics, P.O. Box 655, West Milford, NJ 07480, USA (Phone 201-728-0188, FAX 201-728-0735) for $199. > 4. Simple annotation before printing - bond angles, lengths, torsions etc Nice idea but not trivial. We recommend annoting an image (bitmap or drawing) in a seperate Windows app. To get a complete list of internal coordinates, do an MM+ single point with a log file and PrintLevel=1 in chem.ini (or after running script command "mechanics-print-level 1"). > 6. Option to save MO's so that recalc is not neccesary later. This is addressed to some extent in the SGI version. The grid of points used to generate the orbital isosurface can be saved to a file. > 7. Option to set a damping factor in minimizations. I don't know what you mean by this? Are you talking about geometry optimization or the damping of SCF iterations as a convergence enhancing technique? > 8. 3D orbital plots and densities on PC's. What if users want the pictures, > no matter how long it takes ? Hyperchem authors perhaps underestimated the > informative content and general coolness of these. We'd love this too, but it's a lot of software coding for us that is done by the OS and hardware on the SGI. Some customers are exploring using Autodesk's 3D Studio for this (but I'm embarrased that I've forgotten who!)... > 10. A Crippling Omission. Where did the "only up to Cl" limitation on > semi-empirical calculation come from ? The whole periodic table is available > for EH calcs, and most of the p-block for PM3 and AM1. Adding these elements > should do nothing to the speed or to the memory requirements, as far as I can > see (all the methods are valence shell only). I would be very interested to > learn why exactly this decision was taken. It just seems very strange. Unlike ab initio, where linear combinations of "1s" gaussians are used to created the 2s, 3s, etc., using Slaters in semi-empirical calculations requires writing more code for integrals (and their derivatives) for each increase in principal quantum number. Since our code is all new (in C) this requires significant effort in development and testing, so that's why there is the limitation. It's coming though... > 11. Please don't force users to rename a parameter file for any of the > methods. Let them fool around freely with parameter sets, in files of their > choice, selectable at run-time. You must be referring to quantum mechanics, because we have made it very easy to add or change molecular mechanics parameters. I guess we didn't think people played with semi-empirical parameters as much... (PM3 was a late addition.) > 13. A Pretty Serious Omission. > Offer, via DDE, data somewhat closer to the computing methods : > a) gradient components > b) the capability to draw simple graphics, like arrows, lines etc > c) orbital coefficients > This would probably spawn the most successfull and prolific user group in > this spiral arm of the galaxy. I agree with a) and c), but for b) find a graphics progam that talks DDE and has a macro language. It could tell HyperChem to copy the image, paste it and then edit it. Thanks for the feedback! Graham ------------ Graham Hurst Hypercube Inc, 7-419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 internet: hurst@hyper.com ________________________________________________________________________ From creon%nas.nasa.gov@autodesk Thu Sep 24 20:12:47 1992 Date: Thu, 24 Sep 92 16:50:19 -0700 From: creon%nas.nasa.gov@autodesk (M. Creon Levit) To: hyperchem@autodesk Subject: Hyperchem and the "protein folding potential" of Crippen and Snow Reply-To: creon%nas.nasa.gov@autodesk I would like to use hyperchem for some protein folding investigation and secondary structure predictions from sequence data. The potential function that is of most interest to me is the "protein folding potential" of Crippen and Snow. It treats each amino acid as a single "atom", and has special n+1, n+2, n+3, and non-bonded force function coefficients on a per-amino acid (ntuple) basis. The work is referenced at the beginning of the Hyperchem "getting started" manual, though it is not mentioned again in the manuals as far as I can tell. It seems like it might be possible to configure hyperchem to use this potential by some series of hacks to rules files. Has anyone done this? Is there a better way? Will this potential function be available in a future version of hyperchem? What other approaches would you recomend for protein folding / structure prediction where only the amino acid sequence is known, and very little is available about the secondary structure? Are their any other software packages, based on, say, AI techniques that could help me as well? Thanks a lot (reply to the list, I guess, since this may be of general interest) Creon Levit Mail Stop T045-1 NASA Ames Research Center Moffett Field, California 94035-1000 USA (415)-604-4403 (phone) (415)-604-3957 (FAX) creon@nas.nasa.gov (Internet) ________________________________________________________________________ From kr%cup.portal.com@autodesk Mon Sep 28 17:37:11 1992 From: kr%cup.portal.com@autodesk To: hyperchem@autodesk Subject: Speed Date: Sat, 26 Sep 92 04:09:00 PDT On 9/21/92 hurst@hyper.hyper.com (Graham Hurst) has posted on sci.chem: (excerpted:) ->I must say that this appraisal is at odds with what I have heard from ->others who have used it, other than the blanket complaint that ->computational chemistry is numerically demanding (and hence "slow"). ->Are your judgements arrived at in comparision with other modelling ->software? I would be interested to know how you were evaluating it. ->Would you be kind enough to email details to me? (Post if you prefer, ->but that might generate more advertising! ;-) Q.3 (9/26/92) Slower Energy Minimization I am using the two programs HyperChem and PolyGraf on the same SGI workstation, so I can compare their performance a little bit. Though I have not investigated this by benchmarking, it seems to me that energy minimizations in HyperChem regularly takes up a lot more time than on PolyGraf. Are there any efforts at HyperChem to optimize the minimization-algorithms or is that rather low on the priority list ? Q.4 (9/26/92) Slower Rotation of Selection vs. Everything Also, an intriguing phenomenon I have noticed is that in HyperChem the rotation of the complete workspace is much smoother and more jerk-free than the rotation of only a few (or one) selected molecule(s) while leaving the other molecules stationary. What is the reason for this effect ? It seems to me that rotating only a selection instead of everything should be faster and smoother, if anything. This must be a mistake in the algorithms, because I can't see any fundamental reason why such an effect would have to appear if there are no programming errors. Thank you for an answer. Markus ------- End - of - Message ----------------------- (responsibility for my personal opinions is mine) Richard Stallman : "The more society becomes information instead of material objects, this question of whether a person can copy things will make the difference between a world of universal prosperity or a world of constant rat race." __________________________________________________ kr@cup.portal.com IIIII M M M M Markus Krummenacker I MM MM MM MM (working for I M M M M M M The Institute for I M M M M Molecular Manufacturing) I M M M M IIIII M M M M ________________________________________________________________________ From kr%cup.portal.com@autodesk Mon Sep 28 17:37:22 1992 From: kr%cup.portal.com@autodesk To: hyperchem@autodesk Subject: Re: Speed Date: Sun, 27 Sep 92 05:32:06 PDT Almost instantly, I received the following answer which probably makes a lot of sense. Thank you very much. ----------------------- Message-Id: <9209260750.AA13659@nova.unix.portal.com> To: kr@cup.portal.com Subject: Re: 3D modelling software Newsgroups: sci.chem In-Reply-To: <66633@cup.portal.com> Organization: Harvey Mudd College, Claremont, CA 91711 Date: Sat, 26 Sep 92 0:47:37 PDT From: ebrandt@jarthur.Claremont.EDU Sender: ebrandt@jarthur.Claremont.EDU In article <66633@cup.portal.com> you write: >Q.4 (9/26/92) Slower Rotation of Selection vs. Everything >Also, an intriguing phenomenon I have noticed is that in HyperChem the rotation >of the complete workspace is much smoother and more jerk-free than the rotation >of only a few (or one) selected molecule(s) while leaving the other molecule This is probably an artifact of the fact that you're using high-performance graphics hardware, namely SGI's. The SGI viewing model, and presumably others, provides a way to change the global viewing transformation -- e.g. rotate everything. This is very fast, because it all happens in hardware. To rotate only one thing requires that the software alter a portion of the display list that it's feeding into the graphics engine. The oddity is really not that rotating a selection is jerkier but that rotating the workspace is smoother, a consequence of the special-case speedup allowed by the graphics hardware. Eli ebrandt@jarthur.claremont.edu ________________________________________________________________________ From autodesk!hyper!slee Mon Sep 28 17:42:29 1992 From: autodesk!hyper!slee (Thomas Slee) Subject: Re: Hyperchem and the "protein folding potential" of Crippen and Snow To: hyperchem@autodesk (HyperChem e-mail group) Date: Fri, 25 Sep 92 10:37:37 EDT A few comments on Creon Levit's e-mail: > > I would like to use hyperchem for some protein folding investigation and > secondary structure predictions from sequence data. The potential function > that is of most interest to me is the "protein folding potential" of > Crippen and Snow. It treats each amino acid as a single "atom", and has > special n+1, n+2, n+3, and non-bonded force function coefficients on a > per-amino acid (ntuple) basis. > > The work is referenced at the beginning of the Hyperchem "getting started" > manual, though it is not mentioned again in the manuals as far as I can > tell. I don't think it is referenced in a context that suggests you can use this potential function in HyperChem. > > It seems like it might be possible to configure hyperchem to use this > potential by some series of hacks to rules files. > It is easy to modify parameters and rules in HyperChem, but in the present release the functional form of the force field is not editable. You would have to be able to cast the Crippen and Snow potential in terms of one of the existing potentials. I don't know if this is possible or not -- seems to me unlikely, unfortunately, but I'm not familiar with the work. > > Will this potential function be available in a future version of hyperchem? > It won't be in the next release. As for after that, nothing is settled yet. I imagine you would like it and will add it to our wishlist. > What other approaches would you recomend for protein folding / structure > prediction where only the amino acid sequence is known, and very little is > available about the secondary structure? Are their any other software > packages, based on, say, AI techniques that could help me as well? > There are the homology-based approaches a la Chou and Fastman, that seem to work well for well-defined structural regions such as alpha helix and beta sheet regions, but seem to fail pretty consistently (from the little I know) in the less well-defined turns and intermediate regions. I don't know about software for this, I am afraid. > Thanks a lot (reply to the list, I guess, since this may be of general > interest) > > Creon Levit > > Mail Stop T045-1 > NASA Ames Research Center > Moffett Field, California 94035-1000 > USA > > (415)-604-4403 (phone) > (415)-604-3957 (FAX) > creon@nas.nasa.gov (Internet) > -- Tom Slee Hypercube, Inc., #7-419 Phillip St., Waterloo, Ont. N2L 3X2 Internet: slee@hyper.com Tel. (519) 725-4040 ________________________________________________________________________ From gedeck%pctc.chemie.uni-erlangen.de@autodesk Thu Oct 22 16:24:07 1992 Date: Thu, 22 Oct 1992 21:22:13 +0200 From: gedeck%pctc.chemie.uni-erlangen.de@autodesk (Peter Gedeck) To: hyperchem@autodesk Subject: mm+ - forcefield parameter for aryl-CN Hello, I tried to optimize benzonitrile using mm+, but without success. The cyano-group turned out to be not linear but the angle C(aryl)-C-N was optimized to 123 degrees. I looked up, if I could find a parameter-set for ca-c2-n1 but didn't find any. Did anyone know parameters for this bond-angle or would it be appropriate to use the given parameter for c4-c2-n1 ? Please respond directly to me, I will summarize to the list. Peter Gedeck Inst.f.Physikalische Chemie I Egerlandstrasse 3 8520 Erlangen Germany Tel: 09131 - 85 7335 Fax: 09131 - 85 8307 E-Mail: gedeck@pctc.chemie.uni-erlangen.de ________________________________________________________________________ From gedeck%pctc.chemie.uni-erlangen.de@autodesk Fri Oct 30 02:23:02 1992 Date: Thu, 29 Oct 1992 17:40:05 +0100 From: gedeck%pctc.chemie.uni-erlangen.de@autodesk (Peter Gedeck) To: hyperchem@autodesk Subject: mm+ - forcefield parameter Hello, Recently I posted a request on the ca-c2-n1 bending parameters for mm+. Tom Slee from autodesk replied with the following letter: There is a bug in the default scheme (in the Windows release only; it was fixed in the SGI release) for MM+ default parameters around sp carbons that sets their default angle to 120 degrees. This bug can be fixed by including an extra line in your mmpben.txt parameter file with the following settings: ** c2 ** 0.400 180.000 0.000 0.000 bugfix This parameter will then be used instead of the hard-wired default parameter settings for cases where no explicit parameter is set: (see the Computational Chemistry manual, Theory and Methods section, p 50 for a full description of how this default scheme works), and will give reasonable angles about sp carbons. The force constant given here is somewhat higher than that of the c4-c2-n1 parameter. I don't necessarily recommend it--I don't know what the best value is for the cyano case--I just took it from one of the other sp carbon bends. My guess is that as there is rarely any reason for cyano groups to be anything other than linear (not being involved in strained systems, for instance) it won't make much difference what you put for the force constant as long as it is not too small: the angle will come out at about 180 degrees anyway, with no significant contribution to the steric energy. Using this force-constant everything works fine. But I just found another problem with mm+. This time it happened calculating dimethylamino-benzene. Geometry-optimization makes the dimethylamino- group perpendicular to the benzene-ring, whereas AM1-calculations with Mopac give a dihedral of 16 degree. This difference seems to be true to steric-effect. Did anyone derived force-constants for this special case, to take the electronic interaction between the lone-pair of the nitrogen and the pi-system of the benzene-ring? Any help would be welcome Peter Gedeck Inst.f.Physikalische Chemie I Egerlandstrasse 3 8520 Erlangen Germany Tel: 09131 - 85 7335 Fax: 09131 - 85 8307 E-Mail: gedeck@pctc.chemie.uni-erlangen.de ________________________________________________________________________ From DSMITH%uoft02.utoledo.edu@autodesk Fri Oct 30 02:23:12 1992 Date: 29 Oct 1992 17:36:32 -0400 (EDT) From: "DR. DOUGLAS A. SMITH, UNIVERSITY OF TOLEDO" Subject: Re: mm+ - forcefield parameter To: gedeck@pctc.chemie.uni-erlangen.de Cc: hyperchem@autodesk >Using this force-constant everything works fine. But I just found >another problem with mm+. This time it happened calculating >dimethylamino-benzene. Geometry-optimization makes the dimethylamino- >group perpendicular to the benzene-ring, whereas AM1-calculations >with Mopac give a dihedral of 16 degree. This difference seems to be >true to steric-effect. Did anyone derived force-constants for this >special case, to take the electronic interaction between the lone-pair >of the nitrogen and the pi-system of the benzene-ring? We recently published a comparative study of semiempirical, ab initio and experimental data on aromatic amines, with particular attention on the pyramidalization of the amino nitrogen. You can find this work in: J. Comp. Chem. 1992, 13, 640-50 or you can request reprints from me. We found the pyramidalization of the amine nitrogen in dimethylaminobenzene (N,N-dimethylaniline), measures as the angle between the ring and the Me-N-Me plane, to be as follows: MNDO AM1 PM3 AbInitio Exptl 89.9 46.4 33.6 NA 27.0 so I don't know where your value of 16 comes from. 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 autodesk!kenyon.edu!batt@fernwood.mpk.ca.us Fri Oct 30 15:25:52 1992 Date: Fri, 30 Oct 1992 14:35:32 EST From: "R. BATT" To: HYPERCHEM@autodesk.com Subject: semiempirical geometry constraints Is there any way to set a geometry constraint, e.g., bond length, while doing a geometry optimization using a semi-empirical method? Russell Batt Kenyon College Gambier, OH 43022 BATT@KENYON.EDU ________________________________________________________________________ From autodesk@fernwood.mpk.ca.us Thu Nov 26 16:46:44 1992 Date: Thu, 26 Nov 92 15:09:44 EST From: ranck@albert.etown.edu (John P. Ranck) To: hyperchem@autodesk.com Subject: Hyperchem input to Gaussian-9x As part of the activities of the MoleCVUE Consortium [Molecular Computation and Visualization in Undergraduate Education], I have recently written a program to prepare Gaussian 90 input files for the molecule currently in the Hyperchem display. The program is accessed by script from within Hyperchem and the resulting Gaussian input file may be further edited from within Hyperchem. You may obtain this program by ftp: host: vax.etown.edu username: molecvue password: molecvue You will need the 5 files in the GAUSSCOM directory. The description of the program is contained in the GAUSSCOM.DOC file as follows. Please note especially the caveats, lack of warranty, etc. _______________________________________________________________________________ The program GAUSSCOM was written as a programming exercise to test and explore: (1) the Borland GUI and the Borland Resources Workshop; (2) object oriented programming in Borland Turbo Pascal for Windows; (3) communication with Autodesk HyperChem using scripts and DDE. This program is executed by it calling from a script (GAUSSCOM.SCR) within HyperChem. The program uses the molecule(s) currently in the HyperChem display to create an input file (.COM) file for Gaussian 9X according to the options selected. Gaussian 9X is a large program which has many user selectable options. This program was written to provide easy input to Gaussian 9X for undergraduate instruction in computational chemistry and provides for only a few of the more commonly used options. The program does allow the user to further edit the input (.COM) file after it has been created. N.B. This program blindly accepts whatever options the user has selected and does no checking for consistency with what Gaussian 90 allows, e.g., you may select both optimization and vibrational frequencies -- a combination which Gaussian 90 disallows as meaningless. Note also, that although the Z-matrix is constructed in a mathematically logical and consistent manner, it may not be constructed in the way that is most obvious to chemists. ----------------------------------------------------- The following files constitute the GAUSSCOM program: GAUSSCOM.DOC - this file. GAUSSCOM.SCR - a script file to be called by HyperChem. GAUSSCOM.EXE - the executable file called by the script. CHEM.SCR - file to add an entry to the HyperChem Scritps menu. BWCC.DLL - dynamically linked library for Borland interface. The .EXE file and the .SCR files should be copied into the HyperChem directory. The CHEM.SCR file is the one that adds entries to the Script menu. If you already have such a file, you should edit one of the entries in accord with the CHEM.SCR file provided here. Finally, copy BWCC.DLL into your ...\WINDOWS\SYSTEM directory ----------------------------------------------------- This program is provided as is and without any warranties. You may use, copy, and distribute it freely for any instructional or non-commercial purposes. Please communicate any problems and/or suggestions to: John P. Ranck Internet: RANCK@VAX.ETOWN.EDU Department of Chemistry Voice: 717-361-1315 Elizabethtown College FAX: 717-361-1207 Elizabethtown, PA 17022 ________________________________________________________________________ From autodesk!msu.edu!16990KMR@fernwood.mpk.ca.us Sat Nov 21 15:33:26 1992 Date: Sat, 21 Nov 92 14:28 EST To: hyperchem@autodesk.com From: "Vincent.Kramer" <16990KMR@msu.edu> Subject: 3d searches and receptor "fit" programs I would like to obtain information on any commercially available 3D searching programs for the identification of specific pharmacophores. Also, I would like to obtain a program for evaluating the "fit" of a compound to a protein receptor, e.g. DOCK. I am just starting out in this field and would appreciate any information. Thank you. Vincent J. Kramer