From jkl@ccl.net Fri Feb 17 01:40:19 1995 Received: for jkl@ccl.net by www.ccl.net (8.6.9/930601.1506) id BAA11250; Fri, 17 Feb 1995 01:33:28 -0500 Date: Fri, 17 Feb 1995 01:33:28 -0500 From: Jan Labanowski Message-Id: <199502170633.BAA11250@www.ccl.net> To: chemistry@ccl.net Subject: Re: CCL:Help for Mopac In-Reply-To: Mail from 'servant@cassis-gw.univ-brest.fr (Gwenael Servant)' dated: Thu, 16 Feb 95 22:15:29 +0100 Cc: jkl@ccl.net servant@cassis-gw.univ-brest.fr (Gwenael Servant) writes: > I have tried to get MOPAC for DOS and for LINUX by www.ccl.net > but the DOS zipped version has a bug (I can't unzip it) > and the Linux version doesn't work on my system. I checked the DOS version in /pub/chemistry/software/MS-DOS/mopac_for_dos/mopacdos.zip on anon.ftp www.ccl.net It unzips without the error. Did you use a binary mode of ftp when downloading? As to bug in LINUX verision (in /pub/chemistry/software/LINUX/mopac7 on www.ccl.net), I did not test it much, but it executes and runs examples under LINUX on my pentium (which still has the bad chip inside...). Jan jkl@ccl.net -- Dr. Jan K. Labanowski, Senior Research/Supercomputer Scientist/Specialist, etc. Ohio Supercomputer Center, 1224 Kinnear Rd, Columbus, OH 43212-1163 ph:(614)-292-9279, FAX:(614)-292-7168, E-mail: jkl@ccl.net JKL@OHSTPY.BITNET From thep@risc1.lrm.fi.cnr.it Fri Feb 17 04:40:21 1995 Received: from risc1.lrm.fi.cnr.it for thep@risc1.lrm.fi.cnr.it by www.ccl.net (8.6.9/930601.1506) id EAA12840; Fri, 17 Feb 1995 04:35:15 -0500 Received: by risc1.lrm.fi.cnr.it (AIX 3.2/UCB 5.64/4.03) id AA18137; Fri, 17 Feb 1995 10:25:07 +0100 From: thep@risc1.lrm.fi.cnr.it (Pornthep Sompornpisut) Message-Id: <9502170925.AA18137@risc1.lrm.fi.cnr.it> Subject: help for Molscript To: chemistry@ccl.net Date: Fri, 17 Feb 1995 10:25:06 +0100 (NFT) X-Mailer: ELM [version 2.4 PL23] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 140 Dear Colleagues, Could anyone inform me how to get the program MOLSCRIPT? Is the program available via ftp? Thanks in advance Pornthep From gedeck@pctc.chemie.uni-erlangen.de Fri Feb 17 07:41:16 1995 Received: from faui45.informatik.uni-erlangen.de for gedeck@pctc.chemie.uni-erlangen.de by www.ccl.net (8.6.9/930601.1506) id HAA13740; Fri, 17 Feb 1995 07:10:02 -0500 Received: from faui43.informatik.uni-erlangen.de by uni-erlangen.de with SMTP; id AA13312 (5.65c-6/7.3w-FAU); Fri, 17 Feb 1995 13:08:35 +0100 Received: from pctc.chemie.uni-erlangen.de by immd4.informatik.uni-erlangen.de with SMTP; id AA04578 (5.65c-6/7.3m-FAU); Fri, 17 Feb 1995 13:08:33 +0100 Received: from pc1.chemie.uni-erlangen.de by pctc.chemie.uni-erlangen.de (AIX 3.2/UCB 5.64/4.03) id AA22447; Fri, 17 Feb 1995 13:08:23 +0100 Received: by pc1.chemie.uni-erlangen.de (AIX 3.2/UCB 5.64/4.03) id AA20234; Fri, 17 Feb 1995 13:08:21 +0100 Date: Fri, 17 Feb 1995 13:08:16 +0100 (NFT) From: Peter Gedeck To: chemistry@ccl.net Cc: Peter Gedeck Subject: Announce: MolEdit 1.0 Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hi, I thought it's time to announce the new release of my program MolEdit. This is an editor for the MS-DOS system, below you will find a description of its features. MolEdit is freeware, further information is available at the WWW page: http://pctc.chemie.uni-erlangen.de/~gedeck/moledit.html or download the program from ftp://pctc.chemie.uni-erlangen.de/pub/moled If you download and use it, send me a note, so that I can keep you informed about bug-fixes. Also, please read the copyright in the readme-file. Peter --------------------------------------------------------------------------- MolEdit 1.0 *********** Editor for Mopac-Files ====================== Introduction ============ MolEdit is an editor specially designed to prepare input-files for quantum-chemical programs. During the construction of an input-file it is always possible to check the correctness of the defined structure. In case of errors MolEdit provides informative error messages for the implemented input formats. In addition the combination of the editor with a molecule-viewer allows to examine the defined geometrical structure. This is especially valuable for the creation of z-matrices. The molecule-viewer allows to rotate and to translate the molecule. The structure can be displayed in five different ways: o Wire-frame o Ball-and-Stick o Dot-clouds o Grid o Spheres The combination of an editor with a molecule-viewer will make it very easy for beginning students to learn to write z-matrices. Other features included in MolEdit are: o Formatting input-files o Converting between different formats (Mopac <=> GIP) o Deleting Dummy-Atoms o Inserting and deleting of atoms (while keeping the references correct). In the current version MolEdit fully supports several file formats itself. These are: o The Mopac file format, which is used by the semiempirical program packages Ampac, Mopac and Vamp. o The GIP (geometry-interchange-program) format o The Xmol format Other file formats, among those the Gaussian-file formats are accessible through the conversion utility Babel. MolEdit provides a interface that simplifies the usage of Babel. MolEdit is also able to load, edit and save UNIX-Files. The is useful, if UNIX-filesystems are mounted via NFS. MolEdit is a DOS-application written for the IBM-PC-AT and needs a VGA or EGA-card, either monochrome or color. A coprocessor is recommended but not necessary. In case the real-mode-version of the program is used, either the use of expanded-memory or of a disk-cache is recommended. The program was tested with monochrome-VGA-cards and should adjust itself to this mode. ------------------------------------------------------------------------- Peter Gedeck Inst. f. Physikalische Chemie I Egerlandstrasse 3 91058 Erlangen Germany Tel: ++9131 - 85 7335 Fax: ++9131 - 85 8307 E-Mail: gedeck@pctc.chemie.uni-erlangen.de WWW: http://pctc.chemie.uni-erlangen.de/~gedeck/gedeck.html From shubin@email.unc.edu Fri Feb 17 10:40:36 1995 Received: from email.unc.edu for shubin@email.unc.edu by www.ccl.net (8.6.9/930601.1506) id KAA16453; Fri, 17 Feb 1995 10:00:25 -0500 Received: by email.unc.edu (AIX 3.2/UCB 5.64/4.03) id AA179308; Fri, 17 Feb 1995 10:00:19 -0500 Date: Fri, 17 Feb 1995 10:00:19 -0500 (EST) From: Shubin Liu X-Sender: shubin@isisa.oit.unc.edu To: chemistry@ccl.net Subject: CCL:summary:Protein Structure Prediction Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear all CCLers: A few days ago I inquired here the status and methods of protein structure prediction. I received many responses the details of which are enclosed. Special thanks are due to all the responsers, namely: John M. Sauder Eric Bittner Jeffrey L. Nauss Kaiqi Chen J. Mike Sauder Lin Dawei William T. Winter Eric Bauer Stanley Krystek Bob Funchess Marketa Zvelebil Bill Tivol Stanley Krystek Shubin ............................................................................. Shubin Liu Department of Chemistry Email: shubin@email.unc.edu University of North Carolina sliu@mulliken.chem.unc.edu Chapel Hill, NC 27599-3290 Tel : (919) 962-0150(O) ............................................................................. ----------------------------------------------------------------------------- From: John Michael Sauder I saw your posting on the biophysics newsgroup. It wasn't clear whether you were talking about predicting secondary structure or tertiary structure. There are algorithms that predict possible secondary structure (alpha helices, beta strands) based on the primary structure (the amino acid sequence). These algorithms are, of course, not 100% correct, but can usually give a general idea of what secondary structural elements might be present in a protein. This method can be coupled with a comparison of proteins with known structure. If their amino acid sequences are similar, their secondary structure will most likely also be similar. Predicting the tertiary structure (how the secondary structure fits together to form the final folded structure) is much more complex. Two methods are possible: (1) Compare the amino acid sequence with other proteins whose tertiary structures have been solved by x-ray crystallography or NMR distance geometry calculations. If the primary sequences are similar (at least 40-50% homologous), an attempt can be made to "build" the unknown protein based on the structure of the known protein. As the homology of the two sequences decreases, the reliability of this method also decreases. If two sequences are, say, 80% homologous, you could be very certain of the predicted structure of your unknown protein. However, if the homology is less than 40-50%, it is often uncertain whether or not your prediction is at all reliable. The second method involves a deep understanding of all the forces involved in keeping a protein in its native (folded) state, such as van der Waals interactions, electrostatic and hydrogen bonding interactions, allowable bond distances, bond angles, and dihedral angles, etc. Parameters representing all of these factors can used to create an energy function which a computer can use to try to predict the protein structure. The computer time required to do the calculations is very large, and the imperfection in the energy function and accompanying parameters makes this method very ineffective. It will probably be decades before this approach is feasible. --------------------------------------------------------------------------- From: Eric Bittner Peter Wolynes and co-workers have devoted a great deal of attention to this problem. Wolynes and Frauenfelder wrote a "Physics Today" article about a year ago on the topic of protein folding (Feb 1994 issue of Phys. Today). Since Peter and Hans are certainly the world experts in this area, I'd suggest communicating directly with either of them. E. R. Bittner Univ of Texas -------------------------------------------------------------------------- From: "Jeffrey L. Nauss" You really need to check the literature first. A simple search should have uncovered several references. There are many algorithms out there to predict secondary structure, Chou-Fasman, Garnier, etc. Some (if not all) have been made into computer programs which are available in the public domain. Now if you are looking for how the secondary structural elements fold into a tertiary structure, that is entering the protein folding problem. Again it has been discussed extensively in the literature with several programs discussed. Don't have any readily available as it is outside my area but I know they exist. Keep in mind that neither the secondary or tertiary structure prediction methods are very accurate. For the secondary prediction methods the best I have seen are about ~70% accurate. Jeff Nauss ---------------------------------------------------------------------------- From: "mark w. dalton" Hi! Yes, there are quite a few programs for secondary structure prediction or proteins. Some are: Antheprot: deleage@ibcp.fr, pilote@ibcp.fr This is a general annoucement of the availability of ANTHEPROT to all academic researchers. ANTHEPROT (ANalyze THE PROTeins) is a package to make protein sequence analysis such as alignment, secondary structure predictions, sites & function detection, physico-chemical profiles, homology search and 3D display of protein structures. This program is now available either for IBM RISC 6000 workstations or IBM PC compatible microcomputers. The main feature of ANTHEPROT is that it is fully interactive within a graphical interface. No particular knowledge about computers is needed and any mole- cular biologist is able to use it. SeqSee - Science( 258:p1369) or ftp canopus.biochem.ualberta.ca A suite of programs for protein sequence analysis ProfileScan- forming a profile for given aligned sequences. by Michael Gribskov gribskov@sdsc.edu (formerly of one of the National Labs in the US). Scrutineer - a protein sequence motif analysis program You also may want to look into Scrutineer you can get it at: netserv@embl.bitnet Gribskov, M., McLachlan, A.D. and Eisenberg, D. (1987) Profile analysis: detection of distantly related proteins. PNAS USA 84, 4355-4358. Gribskov, M., Methods in Enzymology (1990) 183, PP 146-159 URL = http://pscinfo.psc.edu/general/software/cray/c90/profiless/profiless.html PROFILE-SS is a program that combines the optimal local sequence alignment algorithm developed by Michael Watterman and Mark Eggert and the Profile analysis methods of Gribskov into one program that finds the N-best alignments between a database sequence and a profile. You may also want to look at the Protein course that is being worked on I think they may have links to place to find software for this. http://seqnet.dl.ac.uk:8000/vsns-pps/ I have lists of software for 3D prediction and 3D searches also some papers on these topics. What in particular are you trying to study? For 2D predictions almost any decent Genetics software package on any platform does 2D predictions. Good luck! Mark ----------------------------------------------------------------------------- From: Kaiqi Chen As far as I know, there is a book: "Prediction of protein structure and the principles of protein conformation", edited by Gerald D. Fasman. kaiqi chen chen@hbar.rice.edu ----------------------------------------------------------------------------- From: "J. Mike Sauder" I'll do a little searching around, but just a quick search yielded a couple of references you might want to look up. The classic reference is Chou and Fasman. Some simple parameters for estimating helix or beta sheet formation are termed 'Chou-Fasman parameters'. I'm not sure which of the following references is most helpful, since I don't have them handy: Chou & Fasman (1971) Biochemistry 13:211 Chou & Fasman (1978) Advances in Enzymology 47:45-148 I also found a reference to a book by Fasman that you might find very helpful: Fasman (1989) "Prediction of Protein Structure and Principles of Protein Conformation", Plenum, New York. Hope this is helpful for a start. I'll also look around and see if I can find any places that have algorithms or programs that actually do secondary structure prediction. -- Mike ----------------------------------------------------------------------------- From: Lin Dawei Secondary prediction has a long history, I think. The most famous one is Chou-Fosman methods. You may already know that. It is in fact a statistic method. Up to now, there is many method to predict secondary structure of protein, like Garner method, neutral network method etc. But now the precision of all this methods less than 65% for 3-states prediction (alpha-helix, beta-sheet, random coil). So secondary prediction is realy a tough work to do. Now people also use multiple alignment method to improve the precision, but it also seem have a lot work to do. TiBS and Current opinions in biology are two very good journal to beginners. For the folding problem, which I am also interested in, is not a mature field I think. Now, there is no method can predict protein folding just according to its primary sequence. But people have already develop many excellent methods to know better about it. You can also read the above two journals I recommend for recent advantence. --------------------------------------------------------------------------- From: "William T. Winter" look up the published work of Gerald Fasman in particulra Fasman and Chou. --------------------------------------------------------------------------- From: Erich Bauer anyway, it would be fine to know about your background: there are millions of billions of articels/books etc. on that suject, but no clear answer to the proteinfolding problem - and sec.str.pred. is part of this - has been given yet. > basicly there are two approaches: heuristics, such as neural net procedures with an average accuracy up to 70% ( Rost/Sander) and no-knowledgebased, very time consuming methods. if you want, i can point you out some more literature but essentially people think it works as follow: after transcription/processing etc. the protein folds a) in vivo f.i. protected from agglomeration by chaperons b) in vitro spontanously from a random coil very fast into a molten globe state which has lots of secondary structure. secondary structure formation is the initial part of folding, basicly driven by hydrophobic interactions, the molten globe state contains already a lot of regulariteis ( i.e. sec.str. exactly as the later natural state does.) in that sense you may call this a fundamental ( building) block, although nothing is decided. this process is very fast and probably follows a pathway, i.e. a cascade of well defined yet flxible intermediate states. from there it takes a time scale of seconds to refold into the native state, which is the global or a very good local optimum. there is, however several regions of high and others of binary uncertainity, leaving the molecule flexible enough to expell water and resort the overall configuration of such sec.str. elements. maybe long range interactions, forming in the later stages of folding are essential for the formation of the final 3D ( i.e. tertiary) structure. these are the reason ( probably ) why prediction over 70% is tough. send an email to PredictProtein@EMBL-Heidelberg.DE with a cont. like this: predict secondary structure concise result #An2_119 RNKFSKMTDNKW..... and you will get the answer to our sequence. open for further questions erich ------------------------------------------------------------------------------ From: Stanley Krystek The area you comment on has been of great concern to many of us who are chemists, biochemists or computational chemists. There have been many empirical and even more more theoretical approaches developed for generating a set of rules for understanding protein structure and predicting individual secondary structure elements or tertiary folds. I have recently written reviews which will be published this year in the book Current Protocols in Protein Science. We cover the most common empirical methods for structure prediction. One good reference for this work is the book on protein structure prediction written by Fasman. The book is titled "Prediction of Protein Structure and the Principles of Protein Conformation" G. Fasman ed., Plenum Press 1989. There are many problems in using amino acid sequence alone to predict protein structure. For example, protein folding may involve "chaperones" to help in the folding process. Therefore, the "rules" derived for empirical (or statistical) predictions are not a complete description of the folding process. Likewise we have been unable to develop the theoretical tools that would allow us to sample the folding of a protein using molecular dynamics. Our forcefields still approximate many of the important forces involved in the formation of protein structure (i.e. electrostatics and solvations effects). So the questions you ask are ones for which we strive to answer. The attempts to describe through empirical and explicit forcefields protein structure is ongoing in most chemistry, biochemistry, molecular biology departments as well as most pharmaceutical companies. stan ------------------------------------------------------------------------------- From: Bob Funchess There are many programs designed to do just this sort of thing; for some examples of predictor algorithms you might try the following papers: Holley, L.H. and Karplus, M. Proc. Natl. Acad. Sci. USA 86, 152-156 (1989) Garnier, J.; Oglethorp, D.J.; and Robinson, B. J. Mol. Biol. 120, 97-120 (1978) Chou, P.Y. and Fasman, G.D. Adv. in Enzymology 47, 45-148 (1978) (I happen to have these three references in particular handy because these methods or derivatives of them are used in the Molecular Simulations program QUANTA; there are also any number of other methods.) Bob Funchess ------------------------------------------------------------------------------ From: Stanley Krystek The area you comment on has been of great concern to many of us who are chemists, biochemists or computational chemists. There have been many empirical and even more more theoretical approaches developed for generating a set of rules for understanding protein structure and predicting individual secondary structure elements or tertiary folds. I have recently written reviews which will be published this year in the book Current Protocols in Protein Science. We cover the most common empirical methods for structure prediction. One good reference for this work is the book on protein structure prediction written by Fasman. The book is titled "Prediction of Protein Structure and the Principles of Protein Conformation" G. Fasman ed., Plenum Press 1989. There are many problems in using amino acid sequence alone to predict protein structure. For example, protein folding may involve "chaperones" to help in the folding process. Therefore, the "rules" derived for empirical (or statistical) predictions are not a complete description of the folding process. Likewise we have been unable to develop the theoretical tools that would allow us to sample the folding of a protein using molecular dynamics. Our forcefields still approximate many of the important forces involved in the formation of protein structure (i.e. electrostatics and solvations effects). So the questions you ask are ones for which we strive to answer. The attempts to describe through empirical and explicit forcefields protein structure is ongoing in most chemistry, biochemistry, molecular biology departments as well as most pharmaceutical companies. stan -- Stan Krystek Bristol-Myers Squibb krystek@bms.com From marketa@kestrel.ludwig.ucl.ac.ukFri Feb 17 09:43:16 1995 You should look at publications from the following groups; MJE Sternberg et al. GJB Barton et al. JM Thornton et al. TL Blundell et all W. Taylor et al Fred Cohen et al. Zvelebil et al. And there are many more, who are actively working on solving the problem of foldin, prediction and accuracy of prediction of proteins from sequence. Marketa Zvelebil ------------------------------------------------------------------------- Date: 10 FEB 1995 18:25:22 GMT This is a very active field. I don't have specific referrences, but if no expert in this field responds in this newsgroup, I'd suggest cross-posting to several other newsgroups such as bionet.general, bionet.molec-model (I think I have the name right) and any with "protein" in the title. If you still have no luck, try Biophys J. or Protein Science. Good luck. Yours, Bill Tivol -------------------------------------------------------------------------- From miquel@nepal.udg.es Fri Feb 17 10:41:03 1995 Received: from nepal.udg.es for miquel@nepal.udg.es by www.ccl.net (8.6.9/930601.1506) id JAA16259; Fri, 17 Feb 1995 09:50:27 -0500 Received: by nepal.udg.es (AIX 3.2/UCB 5.64/4.03) id AA17007; Fri, 17 Feb 1995 15:45:35 +0100 Date: Fri, 17 Feb 1995 15:45:35 +0100 From: miquel@nepal.udg.es (Miquel Sol…) Message-Id: <9502171445.AA17007@nepal.udg.es> To: chemistry@ccl.net Subject: Replies to Bader analysis of CO Dear CCLers, Many thanks to all who replied to my question. For all who may be interested in the conditions of the calculations, the r(C-O) bond length in CO molecule was taken as 1.128 A. The results yield a charge on the oxygen atom of -0.0145 au, while the dipole moment is 0.1024 D. For the LiF molecule the r(Li-F) was 1.564 A, the charge on fluorine atom -0.6478 au and the dipole moment -6.4895 D. Further, the value of the density in the bond critical point (bcp) is 0.4829 au in CO and 0.0722 au in LiF. Probably, the main conclusion is that the mere value of the laplacian in the bond critical point is not enough to define ionicity or covalency and that one should look also the shape of the laplacian in larger regions around the bcp, as well as the value of the electron density in the bcp. My post was: >An electronic analysis of the CO and LiF systems using the generalized >density from a QCISD/6-311++G** wavefunction yields a value for the laplacian >of the density of 0.8328 au for CO and 0.7389 au for LiF. Could this mean, >from a Bader analysis point of view, that CO is more ionic than LiF !!?? >Shoud the positive value of the laplacian (charge depletion) mean that the C-O >bond in CO molecule is ionic?. > > If there is enough interest, I shall summarize the responses >that I get on the net. > > Thanks in advance, > > Miquel Sola > Institut de Quimica Computacional > Universitat de Girona > Voice: +34.72.41.83.62 > FAX: +34.72.41.83.61 > World Wide Web: http://stark.udg.es > e-mail: miquel@nepal.udg.es > miquel@stark.udg.es Answers: From: EDGECOMK@QUCDN.QueensU.CA Miquel, A positive value of the Laplacian at a 'bond crt. pt' does not neccessarily mean that the bond is an ionic bond. It simply means that the density is locally depleted. Other types of interactions can give you a positive value of the Laplacian as well.. eg H-bond. The positive values have been correlated with 'closed- shell' interactions which include ionic bonds. Out of curiousity, how does your dipole moment stack up in both cases? Just to get an idea of the general charge distribution. Take care... hope this helps. Ken Edgecombe Dept. of Chem. Queen's Univ. Kingston, ON From: smb@smb.chem.niu.edu (Steven Bachrach) >the laplacian >of the density of 0.8328 au for CO and 0.7389 au for LiF. Could this mean, >from a Bader analysis point of view, that CO is more ionic than LiF !!?? Well, suprising as it may seem, this is probably a correct conclusion! If you calculate the charge on C and O in CO, my recollection is that the values are soemthing like +1.2 and -1.2 respectively. The charge on Li and F in LiF is probably something like +0.95 and -0.95, respectively. So which is more ionic? Finally, I don't think that Bader ever suggested that the values of the laplacian are interpretable in this way. All he indicated was that if the laplacian is positive that this indicates a preponderance of ionic character in the bond and a negative value indicates a mainly covalent character. I actually had a long and heated argument with Bader many years ago about this and just what the laplacian really means. I personally have my doubts about interpreting the laplacian in terms of ionic/covalent character. Steve Steven Bachrach Department of Chemistry Northern Illinois University DeKalb, Il 60115 Phone: (815)753-6863 smb@smb.chem.niu.edu Fax: (815)753-4802 From lohrenz@zinc.chem.ucalgary.ca Thu Feb 16 00:09:37 1995 Dear Miguel, I'm not absolutely sure about this, but I would suspect more negative values for the laplacian at least for CO. Your are right: positive values should indicate a more inionic interaction. I carried out calculations on organolithium compounds, and the laplacian at the bond critical point for the C-Li bond was usually in the area of +0.2, Li-F interactions in these molecules were about +0.3. May be there is a bug in the program? G. Frenking from Marburg pointed me to the fact that the energy density (H(rc)) at the bond critical point is a better measure for the covalency of the respective bond. H(rc) is defined as the sum of kinetic energy density (G(rc)) and potential energy density (V(rc)). For covalent bonds it is always negative, whereas ionic, H-bond and vdW interactions have positive densities! ref. D. Cremer, E. Kraka, Croat. Chem. Acta 1984,57, 1259. Furtheron Cioslowski in Tallahassee calculates degrees of ionicity starting from the Bader integrations (J. Cioslowski, S.T. Mixon, JACS 1991,113,4142. Hope this helps John -- ========================================================================= Dr. John Lohrenz Dept. of Chemistry Phone: (403) 220 3232 University of Calgary FAX: (403) 289 9488 2500 University Drive, N.W. Calgary, Alberta, T2N 4C2 email: lohrenz@zinc.chem.ucalgary.ca Canada ========================================================================= From NEMCC@CUNYVM.CUNY.EDU Thu Feb 16 03:36:38 1995 Your calculation for LiF is of course for an isolated molecule in the gas phase ..... Li(+) is EXTREMELY electrophilic in the gas phase, and would be expected to have almost as great a "pull" on a pair of electrons as a fluorine nucleus, so your result is not chemically surprising at all. CO is another story. Its dipole moment is almost zero: 0.04 Debyes, if I remember correctly, and any calculation should show this. I'll take a look through Bader's recent book on his doctrines, and see if I can come up with an answer ! Check the gas phase electron affinities of Li(+) and F. The electron affinity of a fluorine atom in the gas phase (no stabilisation by hydrogen bonding) is surprisingly low. From jig@qorg.unizar.es Thu Feb 16 09:11:40 1995 Dear Miquel, With all my respect to the laplacian of the density and the Bader analysis, which is the conductivity of a solution of CO in water? And the same question for a solution of LiCl in water. Is the CO dissociated in C(2+) and O(2-) in solution? Sorry for joking with your problem. I know that the true question is why the Bader analysis gives this bizarre result for these molecules, but I cannot resist to give the "experimentalist" point of view, which is often hard to find in the CCL. Best regards, Jose Ignacio -- ******************************************************************************* Dr. Jose Ignacio Garcia-Laureiro Phone : 34-(9)76-350475 Departamento de Quimica Organica Fax : 34-(9)76-567920 Instituto de Ciencia de Materiales de Aragon e-mail: jig@qorg.unizar.es C.S.I.C.-Universidad de Zaragoza jig@msf.unizar.es E-50009 ZARAGOZA (SPAIN) ******************************************************************************* "And all this science I don't understand it's just my job five days a week..." ELTON JOHN - Rocket man ******************************************************************************* From SERGEI@ps1515.chemie.uni-marburg.de Thu Feb 16 10:36:54 1995 Dear Miquel, Although I do not know exactly the reason for such behaviour of the electron density Laplacian in CO molecule, I have some comments to this subject: 1. Positive values of the Laplacian are typical of C-O multiple bonds as well as of C-F and some other bonds which are normally treated as covalent. The most famous example is F2 where the bond is defi- nitely purely covalent. Probably the positive Laplacian can occur in "electron-rich" molecules. 2. The Bader's topological theory does not state that the bonds with the positive Laplacian are ionic ones. Bader distinguishes between covalent and "closed-shell" interactions. The latter include ionic, donor-acceptor, and van der Waals interactions. 3. It is quite incorrect to compare the absolute values of the Lapla- cian in the bonds formed by different atoms (one cannot say whether LiF is more or less ionic than CO) 4. There exists another criterion of covalency suggested by Cremer and Kraka. Namely, a bond is to be considered as covalent, if the elec- tronic energy density is negative. In CO molecule this is the case. I would not like to discuss whether or not the Cremer-Kraka's cri- terion is better than Bader's one. 5. I have repeated you calculation (QCISD/6-311++G**) of CO molecule at C-O separation of 1 angstroem using Gaussian92 followed by Ba- der's EXTREME program and obtained the value of the electron den- sity Laplacian at the (3,-1) critical point as much as +2.837au. Thus, either in my or in you calculation something is wrong. Yours sincerely, Sergei Vyboishchikov FB Chemie der Universitaet Marburg Hans-Meerwein-Strasse D-35032 Marburg, Germany E-Mail: sergei@ps1515.chemie.uni-marburg.de From D.A.Buttar@bath.ac.uk Thu Feb 16 11:12:29 1995 Miquel, If i understand the Bader Analysis correctly a positive value of the Laplacian does indicate ionic character. However the magnitude of the charge density at the bond critical point must also be considered. For LiF i would expect that the charge density is much smaller at the bcp than in the corresponding CO bond and that this indicates an ionic LiF bond whereas the positive value for the CO indicates a strongly polar, covalent bond. Classification of bonds may also be performed using the energy density at the bond critical point, as discussed by Cremer and Gauss, JACS, 107, 1985, p.3800. I would be very interested in any other replies you receive on the subject. Cheers, David. From csonka@iris.inc.bme.hu Thu Feb 16 11:49:02 1995 FTP Hello Miquel, I suppose that you should analyse larger region around the BCP. The analysis of gradient trajectories would help. I think you will find strange behaviour inside the CO electron density near to the critical point. Try acetilene too. Gabor Gabor I. Csonka Budapest University of Technology Tel/FAX: (361) 463.18.35 Inorganic Chemistry Dept. Ch. Bldg csonka@iris.inc.bme.hu H-1111, Bp. Szent Gellert ter 4 http://www.fsz.bme.hu/bme/chemical/csonka.html From dimas@hobbit.quimica.uniovi.es Thu Feb 16 15:26:08 1995 Hola Miquel, Creo que el libro de Bader (Atoms in Molecules. A Quantum Theory) puede serte util para aclarar tu cuestion. Si mal no recuerdo, en el libro se seganala que las propiedades de la densidad elec- tronica en los puntos criticos de enlace, no son directamente comparables para enlaces constiudos por diferentes pares de atomos. Asi, la magnitud de la densidad electronica de enlace para un C-S no puede comparase con el valor correspondiente en un enlace C-C por ejemplo, para poder decir "quien es mas fuerte que quien". (La teoria no puede respaldar ese tipo de afirmaciones, aunque luego el mismo Bader sin ir mas lejos compare distintos enlaces de puente de H en base al valor de rho(critico) de un modo "empirico"). Ademas, la circunstancia de que el enlace C-O, C-S, presente valores positivos de la Laplaciana, del tipo de los compuestos ionicos LiF, tambien es tratada por Bader. Estos enlaces, C-O, son covalente-polares, implicando una parcial transferencia de carga entre los atomos implicados, de resultas del cual la energia cinetica predomina en la region de enlace. De todos modos, que se trata de una interaccion covalente tambien se puede poner de manifiesto mediante un mapa de la Laplaciana sobre la region de enlace. En dicho mapa se observa que la densidad electronica de C y O se deforma claramente en la region de enlace, mientas que para el caso de compuestos ionicos, lo que se observa es que la laplaciana es practicamente esferica ("dos bolitas") en torno a los atomos en cuestion. Supongo tambien, que si has calculado las cargas atomicas de Bader para el caso del CO y el LiF, tambien tendras otro criterio para distinguir ambos casos. Supongo que en el libro de Bader (cap 7, si mal no recuerdo) podras encontrar detalladamente este tipo de discusion, acerca de la caracterizacion de interacciones atomicas de distintos tipos a partir de las propiedades de los puntos criticos de enlace. Asimismo, la densidad de energia en el punto critico puede ser otra magnitud que diferencie ambos enlaces.(Esta tecnica ya no es de Bader, sino de D. Cremer y E. Kraka y esta expuesta en un par de articulos de agno 1984, en JACS y Angewandte..) Bueno, no se si estos comentarios te pueden servir de algo,... Hasta luego,... Dimas Suarez e-mail: dimas@hobbit.quimica.uniovi.es From !rgab@netcom.com Thu Feb 16 17:47:22 1995 Reply to: RE>CCL:Bader analysis of CO Miquel: You should read the section(s) in Richard Bader's book (at least) on this matter. I do not believe that there is any correlation, per se, between the mag- nitude of the Laplacian and any chemical notion of 'covalency/ionicity'. The Laplacian is simply the trace of the Hessian of Rho; at the bond- critical point, you look at it as the sum of the 3 eigenvalues (of that Hessian) and conclude: is the sum of the transverse curvatures greater in magnitude than the longitudinal curvature? You should note, as follows, there are two 'extremes' of behaviour of the charge density in 'interatomic' regions: "shared" interactions (typically between covalently bound atoms) in which the Laplacian is _usually_ negative; "closed shell" interactions (like noble-gas vdW dimers and 'ionic' systems) in which it is _invariably_ positive. In between there are so-called "intermediate" cases, like CO (in fact most C=O and C=S bonds) in which the interaction is "shared" but the Laplacian is +ve. You have to also look at the magnitude of Rho itself which is usually around 10^-1 a.u. for "shared" interactions. The reason for the sign of the Laplacian in these intermediate cases is that the location of the 'bond critical point' is close to the "VSCC" (Valence shell charge concentration) of one of the atoms - see Bader's book. The point is that you have to look at the _whole_ picture; in the Theory of Atoms in Molecules, it is often not useful to draw drastic conclusions from single numbers. Many people are guilty of that and misinter- pretations are legion. Another paper you should look at is Bader and Essen, J. Chem. Phys., 1984. I also have to ask you: although your numbers look OK to me, you should check that those values of Del^2 do in fact correspond to the sum of eigenvalues, if you are using the program "SADD2R" (or "EXTD2R"). E-mail me if you think this is a problem. Richard Bone __________________________________________________________ Richard G. A. Bone, PhD. Computational Chemist Terrapin Technologies, Inc. South San Francisco USA E-mail rgab@trpntech.com From laidig@erebus.mchem.washington.edu Thu Feb 16 18:27:30 1995 Greetings Miquel: You asked about link between the properties of (3,-1) critical points and the character of bonding; specifically about the sign of the Laplacian and the 'ionic character' of bonding in LiF and CO. While its true that the character of chemical bonding can be summarized by the properties at the 'bond critical point' (BCP), I think its unwise to treat the values as magic numbers which tells all about the nature of interaction between two atoms. One must take stock of the entire set of data to get the correct results from these characterizing properties. The rule of thumb is that a low amount of charge AND a positive Laplacian signify an ionic interaction. In the case of LiF vs. CO, the magnitude of the Laplacian at the BCP of CO is more positive, but there is a good deal more electronic charge than typically found in 'ionic bonds' (see Bader and Essen, JCP, 1984). LiF has the small amount of charge typically found in the interatomic region of ionic interactions. LITHIUM FLUORIDE HF/6-311++G** CARBON MONOXIDE HF/6-311++G** RHO 7.0078622807E-02 RHO 5.1842212832E-01 DEL2 7.2110756059E-01 DEL2 1.0834101333E+00 So, the bonding in CO is not the same as that in LiF, 'ionic'. ---Further ramblings--- In fact, the case of CO is particularly interesting as the Laplacian makes clear the seeming contradiction of the different electronegativities of C & O and the small dipole moment for this system. The small dipole moment of CO would lead you to believe that the atoms were roughly electroneutral, based upon the simple idea that the dipole moments is the result of separation of charge. But then how does one reconcile this with the large expected electronegatvities of the two atoms? There is a large transfer of charge from C --> O. But there is also a large opposing polarization of the remaining charge on the carbon atom (unlike what one would expect for spherical ions). It is this large opposing polarization which produces a large dampening of the dipole moments, canceling the contribution from charge transfer, and making the overall dipole nearly zero. This can be seem in the Laplacian map of CO in which the valence shell of carbon is polarized away from the direction of charge transfer, producing a large, relatively diffuse region of charge concentration (see the cover of Bader's paperback to see the Laplacian of CO (blue)). This is why the carbon of CO is so reactive. In addition, the polarization of the carbon valence shell away from oxygen is the reason that the Laplacian is positive at the bond critical point. The Laplacian map of CO shows that the valence shell of O reaches almost to the interatomic surface and that the valence shell of carbon is polarized away from the interatomic surface. So, it is this polarization of C away from O that gives rise to the relatively large value of rho at BCP (a 'sort-of' shared interaction) and the larger, positive magnitude of the Laplacian. Anyway, I think its all pretty interesting... Keith ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Keith E. Laidig Department of Medicinal Chemistry, BG-20, University of Washington, Seattle, WA 98195 E-Mail: laidig@erebus.mchem.washington.edu Phone: (206) 616-2780 or 685-1000 FAX #: (206) 685-3252 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ From ehlers@chem.vu.nl Thu Feb 16 09:28:26 1995 Dear Miquel, instaed of discussing the Laplacian, it is better to compare the energy-density at the bond kritical point (This is the negative K-Value in the Bader-output). For references see: D. Cremer, E. Kraka, Angew. Chem. 8, 614 (1984) (german edition) G. Frenking, D. Cremer, Stucture and Bonding 73, Springer Verlag, Berlin-Heidelberg (1990) W. Koch, G. Frenking, J. Gauss, D. Cremer, J. R. Collins, J. Am. Chem. Soc. 109 5919 (1987). I hope this helps, with best regards, Andreas Ehlers =========================================================================== = Andreas Ehlers - Afdeling Theoretische Chemie, Faculteit Scheikunde = = - Vrije Universiteit Amsterdam = = - De Boelelaan 1083, 1081 HV Amsterdam = = - [ ehlers@iodine.chem.vu.nl = =========================================================================== From mrigank@imtech.ernet.in Fri Feb 17 12:41:30 1995 Received: from sangam.ncst.ernet.in for mrigank@imtech.ernet.in by www.ccl.net (8.6.9/930601.1506) id MAA20173; Fri, 17 Feb 1995 12:24:48 -0500 Received: (from root@localhost) by sangam.ncst.ernet.in (8.6.8.1/8.6.6) with UUCP id WAA29923 for chemistry@ccl.net; Fri, 17 Feb 1995 22:54:13 +0530 Received: from imtech.UUCP by doe.ernet.in (4.1/SMI-4.1-MHS-7.0) id AA27322; Fri, 17 Feb 95 22:14:16+050 Message-Id: <9502171714.AA27322@doe.ernet.in> Received: by imtech.ernet.in (DECUS UUCP w/Smail); Fri, 17 Feb 95 09:38:06 +0530 Date: Fri, 17 Feb 95 09:38:06 +0530 From: Mrigank To: chemistry@ccl.net Subject: RE: CCL:Protein Structure Prediction X-Vms-Mail-To: UUCP%"shubin@email.unc.edu" ==>Recently, I am interested in the theoretical prediction of the secondary ==>structure, that is, folding, of proteins. It seems to me that there ==>should exist some tools, which may be either theoretical or empirical, to ==>predict the folding process of each protein based on its sequence. The ==>fundamental blocks of this process is believed to be the alapha-helix and ==>beta-sheet, which are also believed not to be formed arbitrarily. I found ==>that there seems exist some regularities in the folding of these helixes ==>and sheets, which govern the shape of the secondary structure of the ==>protein. These regularities or rules seem not to be well-recogized or ==>mentioned in references, as of my knowledge whose major is not biology. ==> ==>My questions are, then, is it possible to predict it? if not impossible, ==>what progress has been made in this topic? Has someone mentioned the ==>regularities or rules during the folding process for alapha-helix and ==>beta-sheet as I believed above? ==> Well, secondary str. is not folding. It involves higher oreder stucture of proteins. Yes, there are many methods, but we are still not sure obout what is the information contained within a protein sequence and how to expliot it to get the complete picture. There are several methods as far as secondary structure is concerned. Best rated are one given by Sander's group and SOPM. For higher order strucrure I guess one can only tell about the types of metthod being attepted, but sucess is still far of. Methods which are very succesful include Simulated Anealing based on either MC or MD, Knowledge bases modelling and Genetic Algorithm based modelling. Mrigank ---- /Mrigank \/ Phone +91 172 690557 \ \Institute of Microbial Technology /\ Email: mrigank@imtech.ernet.in / /Sector 39A, \/ FAX: +91 172 690585 \ \Chandigarh 160 014 India. /\ / \//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\// -- When I feed the poor, they call me saint. When I ask why the poors do not have food, they call me communist - Archbishop Camaran ==>Looking forward to hearing from you! Thanks for your attention! ==> ==>Shubin From bbendik@telerama.lm.com Fri Feb 17 17:43:28 1995 Received: from terrazzo.lm.com for bbendik@telerama.lm.com by www.ccl.net (8.6.9/930601.1506) id QAA25898; Fri, 17 Feb 1995 16:48:38 -0500 Received: (from bbendik@localhost) by terrazzo.lm.com (8.6.9/8.6.9) id QAA11228; Fri, 17 Feb 1995 16:48:27 -0500 Date: Fri, 17 Feb 1995 16:48:21 -0500 (EST) From: Beverly Bendiksen X-Sender: bbendik@terrazzo.lm.com To: chemistry@ccl.net Subject: MOPAC 7 ESP Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII I recently compiled MOPAC 7 on my Silicon Graphics Personal Iris 4D35 workstation, intending to do some MNDO ESP charge calculations. However, the test of this subroutine produced ESP charges reported as *nan* or not a number. There were no error messages when I compiled the program, or in its execution. Does anyone know of any modification to the Fortran that need to be made for an SGI workstation for the ESP calculation to run correctly? bev Beverly Bendiksen bbendik@telerama.lm.com Calgon Corp voice(412) 777-8862 Pittsburgh, PA fax(412) 777-8714 From jtgolab@amoco.com Fri Feb 17 19:50:04 1995 Received: from interlock.amoco.com for jtgolab@amoco.com by www.ccl.net (8.6.9/930601.1506) id TAA01071; Fri, 17 Feb 1995 19:19:41 -0500 Received: by interlock.amoco.com id AA11929 (InterLock SMTP Gateway 3.0 for CHEMISTRY@ccl.net); Fri, 17 Feb 1995 18:20:02 -0600 Message-Id: <199502180020.AA11929@interlock.amoco.com> Received: by interlock.amoco.com (Protected-side Proxy Mail Agent-3); Fri, 17 Feb 1995 18:20:02 -0600 Received: by interlock.amoco.com (Protected-side Proxy Mail Agent-2); Fri, 17 Feb 1995 18:20:02 -0600 Received: by interlock.amoco.com (Protected-side Proxy Mail Agent-1); Fri, 17 Feb 1995 18:20:02 -0600 From: jtgolab@amoco.com (Joe Golab) Date: Fri, 17 Feb 1995 08:33:54 -0600 X-Mailer: Z-Mail (3.1.0 22feb94 MediaMail) To: CHEMISTRY@ccl.net Subject: New Iron & Warranties Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 A quick note on purchasing new workstations (if you're like us, it is time to replace old iron with new). Make sure the warranty on the new machines is one you can live with! For example, Silicon Graphics (which makes excellent computer equipment - consider the r8000 [if you can get them to deliver one]) has a standard 90-day warranty BUT it is a "factory exchange" warranty. This means that (depending on the problem) you will have to ship the unit back to the SG factory in CA - without the internal hard disk, memory, floppy, etc. - BEFORE they send out a new system. For a small amount of money (compared to the purchase price) one can obtain a "PARTS CARE" warranty that extends the 90 days to 12 months AND allows for "advance replacement" (which means that you get a new machine before you have to send the old one back). Not all computer makers operate like this. IBM, for example, has the typical 12 month "we do not ship defective products" warranty (which means if there is a problem they are so embarrassed that they have a tech come to you and fix it). And Cray helps you set the equipment up (or used to do so). Can't beat that! Anyway, please take this as the reminder it is intended to be as we all enter the natural cycle of replacing "obsolete" workstations with more advanced ones. Don't assume that the warranty that comes with your new machine is the same as the one that came with your old one - even if the company is the same one. -- :Joe jtgolab@amoco.com (708) 961-7878 +------------------------------------------------------+ | Everywhere is walking distance if you have the time. | | - Steven Wright | +------------------------------------------------------+ From shubin@email.unc.edu Fri Feb 17 10:00:27 1995 Received: from email.unc.edu for shubin@email.unc.edu by www.ccl.net (8.6.9/930601.1506) id KAA16453; Fri, 17 Feb 1995 10:00:25 -0500 Received: by email.unc.edu (AIX 3.2/UCB 5.64/4.03) id AA179308; Fri, 17 Feb 1995 10:00:19 -0500 Date: Fri, 17 Feb 1995 10:00:19 -0500 (EST) From: Shubin Liu X-Sender: shubin@isisa.oit.unc.edu To: chemistry@ccl.net Subject: CCL:summary:Protein Structure Prediction Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Status: RO Dear all CCLers: A few days ago I inquired here the status and methods of protein structure prediction. I received many responses the details of which are enclosed. Special thanks are due to all the responsers, namely: John M. Sauder Eric Bittner Jeffrey L. Nauss Kaiqi Chen J. Mike Sauder Lin Dawei William T. Winter Eric Bauer Stanley Krystek Bob Funchess Marketa Zvelebil Bill Tivol Stanley Krystek Shubin ............................................................................. Shubin Liu Department of Chemistry Email: shubin@email.unc.edu University of North Carolina sliu@mulliken.chem.unc.edu Chapel Hill, NC 27599-3290 Tel : (919) 962-0150(O) ............................................................................. ----------------------------------------------------------------------------- From: John Michael Sauder I saw your posting on the biophysics newsgroup. It wasn't clear whether you were talking about predicting secondary structure or tertiary structure. There are algorithms that predict possible secondary structure (alpha helices, beta strands) based on the primary structure (the amino acid sequence). These algorithms are, of course, not 100% correct, but can usually give a general idea of what secondary structural elements might be present in a protein. This method can be coupled with a comparison of proteins with known structure. If their amino acid sequences are similar, their secondary structure will most likely also be similar. Predicting the tertiary structure (how the secondary structure fits together to form the final folded structure) is much more complex. Two methods are possible: (1) Compare the amino acid sequence with other proteins whose tertiary structures have been solved by x-ray crystallography or NMR distance geometry calculations. If the primary sequences are similar (at least 40-50% homologous), an attempt can be made to "build" the unknown protein based on the structure of the known protein. As the homology of the two sequences decreases, the reliability of this method also decreases. If two sequences are, say, 80% homologous, you could be very certain of the predicted structure of your unknown protein. However, if the homology is less than 40-50%, it is often uncertain whether or not your prediction is at all reliable. The second method involves a deep understanding of all the forces involved in keeping a protein in its native (folded) state, such as van der Waals interactions, electrostatic and hydrogen bonding interactions, allowable bond distances, bond angles, and dihedral angles, etc. Parameters representing all of these factors can used to create an energy function which a computer can use to try to predict the protein structure. The computer time required to do the calculations is very large, and the imperfection in the energy function and accompanying parameters makes this method very ineffective. It will probably be decades before this approach is feasible. --------------------------------------------------------------------------- From: Eric Bittner Peter Wolynes and co-workers have devoted a great deal of attention to this problem. Wolynes and Frauenfelder wrote a "Physics Today" article about a year ago on the topic of protein folding (Feb 1994 issue of Phys. Today). Since Peter and Hans are certainly the world experts in this area, I'd suggest communicating directly with either of them. E. R. Bittner Univ of Texas -------------------------------------------------------------------------- From: "Jeffrey L. Nauss" You really need to check the literature first. A simple search should have uncovered several references. There are many algorithms out there to predict secondary structure, Chou-Fasman, Garnier, etc. Some (if not all) have been made into computer programs which are available in the public domain. Now if you are looking for how the secondary structural elements fold into a tertiary structure, that is entering the protein folding problem. Again it has been discussed extensively in the literature with several programs discussed. Don't have any readily available as it is outside my area but I know they exist. Keep in mind that neither the secondary or tertiary structure prediction methods are very accurate. For the secondary prediction methods the best I have seen are about ~70% accurate. Jeff Nauss ---------------------------------------------------------------------------- From: "mark w. dalton" Hi! Yes, there are quite a few programs for secondary structure prediction or proteins. Some are: Antheprot: deleage@ibcp.fr, pilote@ibcp.fr This is a general annoucement of the availability of ANTHEPROT to all academic researchers. ANTHEPROT (ANalyze THE PROTeins) is a package to make protein sequence analysis such as alignment, secondary structure predictions, sites & function detection, physico-chemical profiles, homology search and 3D display of protein structures. This program is now available either for IBM RISC 6000 workstations or IBM PC compatible microcomputers. The main feature of ANTHEPROT is that it is fully interactive within a graphical interface. No particular knowledge about computers is needed and any mole- cular biologist is able to use it. SeqSee - Science( 258:p1369) or ftp canopus.biochem.ualberta.ca A suite of programs for protein sequence analysis ProfileScan- forming a profile for given aligned sequences. by Michael Gribskov gribskov@sdsc.edu (formerly of one of the National Labs in the US). Scrutineer - a protein sequence motif analysis program You also may want to look into Scrutineer you can get it at: netserv@embl.bitnet Gribskov, M., McLachlan, A.D. and Eisenberg, D. (1987) Profile analysis: detection of distantly related proteins. PNAS USA 84, 4355-4358. Gribskov, M., Methods in Enzymology (1990) 183, PP 146-159 URL = http://pscinfo.psc.edu/general/software/cray/c90/profiless/profiless.html PROFILE-SS is a program that combines the optimal local sequence alignment algorithm developed by Michael Watterman and Mark Eggert and the Profile analysis methods of Gribskov into one program that finds the N-best alignments between a database sequence and a profile. You may also want to look at the Protein course that is being worked on I think they may have links to place to find software for this. http://seqnet.dl.ac.uk:8000/vsns-pps/ I have lists of software for 3D prediction and 3D searches also some papers on these topics. What in particular are you trying to study? For 2D predictions almost any decent Genetics software package on any platform does 2D predictions. Good luck! Mark ----------------------------------------------------------------------------- From: Kaiqi Chen As far as I know, there is a book: "Prediction of protein structure and the principles of protein conformation", edited by Gerald D. Fasman. kaiqi chen chen@hbar.rice.edu ----------------------------------------------------------------------------- From: "J. Mike Sauder" I'll do a little searching around, but just a quick search yielded a couple of references you might want to look up. The classic reference is Chou and Fasman. Some simple parameters for estimating helix or beta sheet formation are termed 'Chou-Fasman parameters'. I'm not sure which of the following references is most helpful, since I don't have them handy: Chou & Fasman (1971) Biochemistry 13:211 Chou & Fasman (1978) Advances in Enzymology 47:45-148 I also found a reference to a book by Fasman that you might find very helpful: Fasman (1989) "Prediction of Protein Structure and Principles of Protein Conformation", Plenum, New York. Hope this is helpful for a start. I'll also look around and see if I can find any places that have algorithms or programs that actually do secondary structure prediction. -- Mike ----------------------------------------------------------------------------- From: Lin Dawei Secondary prediction has a long history, I think. The most famous one is Chou-Fosman methods. You may already know that. It is in fact a statistic method. Up to now, there is many method to predict secondary structure of protein, like Garner method, neutral network method etc. But now the precision of all this methods less than 65% for 3-states prediction (alpha-helix, beta-sheet, random coil). So secondary prediction is realy a tough work to do. Now people also use multiple alignment method to improve the precision, but it also seem have a lot work to do. TiBS and Current opinions in biology are two very good journal to beginners. For the folding problem, which I am also interested in, is not a mature field I think. Now, there is no method can predict protein folding just according to its primary sequence. But people have already develop many excellent methods to know better about it. You can also read the above two journals I recommend for recent advantence. --------------------------------------------------------------------------- From: "William T. Winter" look up the published work of Gerald Fasman in particulra Fasman and Chou. --------------------------------------------------------------------------- From: Erich Bauer anyway, it would be fine to know about your background: there are millions of billions of articels/books etc. on that suject, but no clear answer to the proteinfolding problem - and sec.str.pred. is part of this - has been given yet. > basicly there are two approaches: heuristics, such as neural net procedures with an average accuracy up to 70% ( Rost/Sander) and no-knowledgebased, very time consuming methods. if you want, i can point you out some more literature but essentially people think it works as follow: after transcription/processing etc. the protein folds a) in vivo f.i. protected from agglomeration by chaperons b) in vitro spontanously from a random coil very fast into a molten globe state which has lots of secondary structure. secondary structure formation is the initial part of folding, basicly driven by hydrophobic interactions, the molten globe state contains already a lot of regulariteis ( i.e. sec.str. exactly as the later natural state does.) in that sense you may call this a fundamental ( building) block, although nothing is decided. this process is very fast and probably follows a pathway, i.e. a cascade of well defined yet flxible intermediate states. from there it takes a time scale of seconds to refold into the native state, which is the global or a very good local optimum. there is, however several regions of high and others of binary uncertainity, leaving the molecule flexible enough to expell water and resort the overall configuration of such sec.str. elements. maybe long range interactions, forming in the later stages of folding are essential for the formation of the final 3D ( i.e. tertiary) structure. these are the reason ( probably ) why prediction over 70% is tough. send an email to PredictProtein@EMBL-Heidelberg.DE with a cont. like this: predict secondary structure concise result #An2_119 RNKFSKMTDNKW..... and you will get the answer to our sequence. open for further questions erich ------------------------------------------------------------------------------ From: Stanley Krystek The area you comment on has been of great concern to many of us who are chemists, biochemists or computational chemists. There have been many empirical and even more more theoretical approaches developed for generating a set of rules for understanding protein structure and predicting individual secondary structure elements or tertiary folds. I have recently written reviews which will be published this year in the book Current Protocols in Protein Science. We cover the most common empirical methods for structure prediction. One good reference for this work is the book on protein structure prediction written by Fasman. The book is titled "Prediction of Protein Structure and the Principles of Protein Conformation" G. Fasman ed., Plenum Press 1989. There are many problems in using amino acid sequence alone to predict protein structure. For example, protein folding may involve "chaperones" to help in the folding process. Therefore, the "rules" derived for empirical (or statistical) predictions are not a complete description of the folding process. Likewise we have been unable to develop the theoretical tools that would allow us to sample the folding of a protein using molecular dynamics. Our forcefields still approximate many of the important forces involved in the formation of protein structure (i.e. electrostatics and solvations effects). So the questions you ask are ones for which we strive to answer. The attempts to describe through empirical and explicit forcefields protein structure is ongoing in most chemistry, biochemistry, molecular biology departments as well as most pharmaceutical companies. stan ------------------------------------------------------------------------------- From: Bob Funchess There are many programs designed to do just this sort of thing; for some examples of predictor algorithms you might try the following papers: Holley, L.H. and Karplus, M. Proc. Natl. Acad. Sci. USA 86, 152-156 (1989) Garnier, J.; Oglethorp, D.J.; and Robinson, B. J. Mol. Biol. 120, 97-120 (1978) Chou, P.Y. and Fasman, G.D. Adv. in Enzymology 47, 45-148 (1978) (I happen to have these three references in particular handy because these methods or derivatives of them are used in the Molecular Simulations program QUANTA; there are also any number of other methods.) Bob Funchess ------------------------------------------------------------------------------ From: Stanley Krystek The area you comment on has been of great concern to many of us who are chemists, biochemists or computational chemists. There have been many empirical and even more more theoretical approaches developed for generating a set of rules for understanding protein structure and predicting individual secondary structure elements or tertiary folds. I have recently written reviews which will be published this year in the book Current Protocols in Protein Science. We cover the most common empirical methods for structure prediction. One good reference for this work is the book on protein structure prediction written by Fasman. The book is titled "Prediction of Protein Structure and the Principles of Protein Conformation" G. Fasman ed., Plenum Press 1989. There are many problems in using amino acid sequence alone to predict protein structure. For example, protein folding may involve "chaperones" to help in the folding process. Therefore, the "rules" derived for empirical (or statistical) predictions are not a complete description of the folding process. Likewise we have been unable to develop the theoretical tools that would allow us to sample the folding of a protein using molecular dynamics. Our forcefields still approximate many of the important forces involved in the formation of protein structure (i.e. electrostatics and solvations effects). So the questions you ask are ones for which we strive to answer. The attempts to describe through empirical and explicit forcefields protein structure is ongoing in most chemistry, biochemistry, molecular biology departments as well as most pharmaceutical companies. stan -- Stan Krystek Bristol-Myers Squibb krystek@bms.com >From marketa@kestrel.ludwig.ucl.ac.ukFri Feb 17 09:43:16 1995 You should look at publications from the following groups; MJE Sternberg et al. GJB Barton et al. JM Thornton et al. TL Blundell et all W. Taylor et al Fred Cohen et al. Zvelebil et al. And there are many more, who are actively working on solving the problem of foldin, prediction and accuracy of prediction of proteins from sequence. Marketa Zvelebil ------------------------------------------------------------------------- Date: 10 FEB 1995 18:25:22 GMT This is a very active field. I don't have specific referrences, but if no expert in this field responds in this newsgroup, I'd suggest cross-posting to several other newsgroups such as bionet.general, bionet.molec-model (I think I have the name right) and any with "protein" in the title. If you still have no luck, try Biophys J. or Protein Science. Good luck. Yours, Bill Tivol -------------------------------------------------------------------------- From mrigank@imtech.ernet.in Fri Feb 17 12:24:56 1995 Received: from sangam.ncst.ernet.in for mrigank@imtech.ernet.in by www.ccl.net (8.6.9/930601.1506) id MAA20173; Fri, 17 Feb 1995 12:24:48 -0500 Received: (from root@localhost) by sangam.ncst.ernet.in (8.6.8.1/8.6.6) with UUCP id WAA29923 for chemistry@ccl.net; Fri, 17 Feb 1995 22:54:13 +0530 Received: from imtech.UUCP by doe.ernet.in (4.1/SMI-4.1-MHS-7.0) id AA27322; Fri, 17 Feb 95 22:14:16+050 Message-Id: <9502171714.AA27322@doe.ernet.in> Received: by imtech.ernet.in (DECUS UUCP w/Smail); Fri, 17 Feb 95 09:38:06 +0530 Date: Fri, 17 Feb 95 09:38:06 +0530 From: Mrigank To: chemistry@ccl.net Subject: RE: CCL:Protein Structure Prediction X-Vms-Mail-To: UUCP%"shubin@email.unc.edu" Status: R ==>Recently, I am interested in the theoretical prediction of the secondary ==>structure, that is, folding, of proteins. It seems to me that there ==>should exist some tools, which may be either theoretical or empirical, to ==>predict the folding process of each protein based on its sequence. The ==>fundamental blocks of this process is believed to be the alapha-helix and ==>beta-sheet, which are also believed not to be formed arbitrarily. I found ==>that there seems exist some regularities in the folding of these helixes ==>and sheets, which govern the shape of the secondary structure of the ==>protein. These regularities or rules seem not to be well-recogized or ==>mentioned in references, as of my knowledge whose major is not biology. ==> ==>My questions are, then, is it possible to predict it? if not impossible, ==>what progress has been made in this topic? Has someone mentioned the ==>regularities or rules during the folding process for alapha-helix and ==>beta-sheet as I believed above? ==> Well, secondary str. is not folding. It involves higher oreder stucture of proteins. Yes, there are many methods, but we are still not sure obout what is the information contained within a protein sequence and how to expliot it to get the complete picture. There are several methods as far as secondary structure is concerned. Best rated are one given by Sander's group and SOPM. For higher order strucrure I guess one can only tell about the types of metthod being attepted, but sucess is still far of. Methods which are very succesful include Simulated Anealing based on either MC or MD, Knowledge bases modelling and Genetic Algorithm based modelling. Mrigank ---- /Mrigank \/ Phone +91 172 690557 \ \Institute of Microbial Technology /\ Email: mrigank@imtech.ernet.in / /Sector 39A, \/ FAX: +91 172 690585 \ \Chandigarh 160 014 India. /\ / \//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\//\// -- When I feed the poor, they call me saint. When I ask why the poors do not have food, they call me communist - Archbishop Camaran ==>Looking forward to hearing from you! Thanks for your attention! ==> ==>Shubin From bbendik@telerama.lm.com Fri Feb 17 16:48:41 1995 Received: from terrazzo.lm.com for bbendik@telerama.lm.com by www.ccl.net (8.6.9/930601.1506) id QAA25898; Fri, 17 Feb 1995 16:48:38 -0500 Received: (from bbendik@localhost) by terrazzo.lm.com (8.6.9/8.6.9) id QAA11228; Fri, 17 Feb 1995 16:48:27 -0500 Date: Fri, 17 Feb 1995 16:48:21 -0500 (EST) From: Beverly Bendiksen X-Sender: bbendik@terrazzo.lm.com To: chemistry@ccl.net Subject: MOPAC 7 ESP Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Status: R I recently compiled MOPAC 7 on my Silicon Graphics Personal Iris 4D35 workstation, intending to do some MNDO ESP charge calculations. However, the test of this subroutine produced ESP charges reported as *nan* or not a number. There were no error messages when I compiled the program, or in its execution. Does anyone know of any modification to the Fortran that need to be made for an SGI workstation for the ESP calculation to run correctly? bev Beverly Bendiksen bbendik@telerama.lm.com Calgon Corp voice(412) 777-8862 Pittsburgh, PA fax(412) 777-8714 From jtgolab@amoco.com Fri Feb 17 19:19:43 1995 Received: from interlock.amoco.com for jtgolab@amoco.com by www.ccl.net (8.6.9/930601.1506) id TAA01071; Fri, 17 Feb 1995 19:19:41 -0500 Received: by interlock.amoco.com id AA11929 (InterLock SMTP Gateway 3.0 for CHEMISTRY@ccl.net); Fri, 17 Feb 1995 18:20:02 -0600 Message-Id: <199502180020.AA11929@interlock.amoco.com> Received: by interlock.amoco.com (Protected-side Proxy Mail Agent-3); Fri, 17 Feb 1995 18:20:02 -0600 Received: by interlock.amoco.com (Protected-side Proxy Mail Agent-2); Fri, 17 Feb 1995 18:20:02 -0600 Received: by interlock.amoco.com (Protected-side Proxy Mail Agent-1); Fri, 17 Feb 1995 18:20:02 -0600 From: jtgolab@amoco.com (Joe Golab) Date: Fri, 17 Feb 1995 08:33:54 -0600 X-Mailer: Z-Mail (3.1.0 22feb94 MediaMail) To: CHEMISTRY@ccl.net Subject: New Iron & Warranties Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 Status: R A quick note on purchasing new workstations (if you're like us, it is time to replace old iron with new). Make sure the warranty on the new machines is one you can live with! For example, Silicon Graphics (which makes excellent computer equipment - consider the r8000 [if you can get them to deliver one]) has a standard 90-day warranty BUT it is a "factory exchange" warranty. This means that (depending on the problem) you will have to ship the unit back to the SG factory in CA - without the internal hard disk, memory, floppy, etc. - BEFORE they send out a new system. For a small amount of money (compared to the purchase price) one can obtain a "PARTS CARE" warranty that extends the 90 days to 12 months AND allows for "advance replacement" (which means that you get a new machine before you have to send the old one back). Not all computer makers operate like this. IBM, for example, has the typical 12 month "we do not ship defective products" warranty (which means if there is a problem they are so embarrassed that they have a tech come to you and fix it). And Cray helps you set the equipment up (or used to do so). Can't beat that! Anyway, please take this as the reminder it is intended to be as we all enter the natural cycle of replacing "obsolete" workstations with more advanced ones. Don't assume that the warranty that comes with your new machine is the same as the one that came with your old one - even if the company is the same one. -- :Joe jtgolab@amoco.com (708) 961-7878 +------------------------------------------------------+ | Everywhere is walking distance if you have the time. | | - Steven Wright | +------------------------------------------------------+ From miquel@nepal.udg.es Fri Feb 17 09:52:28 1995 Received: from nepal.udg.es for miquel@nepal.udg.es by www.ccl.net (8.6.9/930601.1506) id JAA16259; Fri, 17 Feb 1995 09:50:27 -0500 Received: by nepal.udg.es (AIX 3.2/UCB 5.64/4.03) id AA17007; Fri, 17 Feb 1995 15:45:35 +0100 Date: Fri, 17 Feb 1995 15:45:35 +0100 From: miquel@nepal.udg.es (Miquel Sol…) Message-Id: <9502171445.AA17007@nepal.udg.es> To: chemistry@ccl.net Subject: Replies to Bader analysis of CO Status: RO Dear CCLers, Many thanks to all who replied to my question. For all who may be interested in the conditions of the calculations, the r(C-O) bond length in CO molecule was taken as 1.128 A. The results yield a charge on the oxygen atom of -0.0145 au, while the dipole moment is 0.1024 D. For the LiF molecule the r(Li-F) was 1.564 A, the charge on fluorine atom -0.6478 au and the dipole moment -6.4895 D. Further, the value of the density in the bond critical point (bcp) is 0.4829 au in CO and 0.0722 au in LiF. Probably, the main conclusion is that the mere value of the laplacian in the bond critical point is not enough to define ionicity or covalency and that one should look also the shape of the laplacian in larger regions around the bcp, as well as the value of the electron density in the bcp. My post was: >An electronic analysis of the CO and LiF systems using the generalized >density from a QCISD/6-311++G** wavefunction yields a value for the laplacian >of the density of 0.8328 au for CO and 0.7389 au for LiF. Could this mean, >from a Bader analysis point of view, that CO is more ionic than LiF !!?? >Shoud the positive value of the laplacian (charge depletion) mean that the C-O >bond in CO molecule is ionic?. > > If there is enough interest, I shall summarize the responses >that I get on the net. > > Thanks in advance, > > Miquel Sola > Institut de Quimica Computacional > Universitat de Girona > Voice: +34.72.41.83.62 > FAX: +34.72.41.83.61 > World Wide Web: http://stark.udg.es > e-mail: miquel@nepal.udg.es > miquel@stark.udg.es Answers: From: EDGECOMK@QUCDN.QueensU.CA Miquel, A positive value of the Laplacian at a 'bond crt. pt' does not neccessarily mean that the bond is an ionic bond. It simply means that the density is locally depleted. Other types of interactions can give you a positive value of the Laplacian as well.. eg H-bond. The positive values have been correlated with 'closed- shell' interactions which include ionic bonds. Out of curiousity, how does your dipole moment stack up in both cases? Just to get an idea of the general charge distribution. Take care... hope this helps. Ken Edgecombe Dept. of Chem. Queen's Univ. Kingston, ON From: smb@smb.chem.niu.edu (Steven Bachrach) >the laplacian >of the density of 0.8328 au for CO and 0.7389 au for LiF. Could this mean, >from a Bader analysis point of view, that CO is more ionic than LiF !!?? Well, suprising as it may seem, this is probably a correct conclusion! If you calculate the charge on C and O in CO, my recollection is that the values are soemthing like +1.2 and -1.2 respectively. The charge on Li and F in LiF is probably something like +0.95 and -0.95, respectively. So which is more ionic? Finally, I don't think that Bader ever suggested that the values of the laplacian are interpretable in this way. All he indicated was that if the laplacian is positive that this indicates a preponderance of ionic character in the bond and a negative value indicates a mainly covalent character. I actually had a long and heated argument with Bader many years ago about this and just what the laplacian really means. I personally have my doubts about interpreting the laplacian in terms of ionic/covalent character. Steve Steven Bachrach Department of Chemistry Northern Illinois University DeKalb, Il 60115 Phone: (815)753-6863 smb@smb.chem.niu.edu Fax: (815)753-4802 >From lohrenz@zinc.chem.ucalgary.ca Thu Feb 16 00:09:37 1995 Dear Miguel, I'm not absolutely sure about this, but I would suspect more negative values for the laplacian at least for CO. Your are right: positive values should indicate a more inionic interaction. I carried out calculations on organolithium compounds, and the laplacian at the bond critical point for the C-Li bond was usually in the area of +0.2, Li-F interactions in these molecules were about +0.3. May be there is a bug in the program? G. Frenking from Marburg pointed me to the fact that the energy density (H(rc)) at the bond critical point is a better measure for the covalency of the respective bond. H(rc) is defined as the sum of kinetic energy density (G(rc)) and potential energy density (V(rc)). For covalent bonds it is always negative, whereas ionic, H-bond and vdW interactions have positive densities! ref. D. Cremer, E. Kraka, Croat. Chem. Acta 1984,57, 1259. Furtheron Cioslowski in Tallahassee calculates degrees of ionicity starting from the Bader integrations (J. Cioslowski, S.T. Mixon, JACS 1991,113,4142. Hope this helps John -- ========================================================================= Dr. John Lohrenz Dept. of Chemistry Phone: (403) 220 3232 University of Calgary FAX: (403) 289 9488 2500 University Drive, N.W. Calgary, Alberta, T2N 4C2 email: lohrenz@zinc.chem.ucalgary.ca Canada ========================================================================= >From NEMCC@CUNYVM.CUNY.EDU Thu Feb 16 03:36:38 1995 Your calculation for LiF is of course for an isolated molecule in the gas phase ..... Li(+) is EXTREMELY electrophilic in the gas phase, and would be expected to have almost as great a "pull" on a pair of electrons as a fluorine nucleus, so your result is not chemically surprising at all. CO is another story. Its dipole moment is almost zero: 0.04 Debyes, if I remember correctly, and any calculation should show this. I'll take a look through Bader's recent book on his doctrines, and see if I can come up with an answer ! Check the gas phase electron affinities of Li(+) and F. The electron affinity of a fluorine atom in the gas phase (no stabilisation by hydrogen bonding) is surprisingly low. >From jig@qorg.unizar.es Thu Feb 16 09:11:40 1995 Dear Miquel, With all my respect to the laplacian of the density and the Bader analysis, which is the conductivity of a solution of CO in water? And the same question for a solution of LiCl in water. Is the CO dissociated in C(2+) and O(2-) in solution? Sorry for joking with your problem. I know that the true question is why the Bader analysis gives this bizarre result for these molecules, but I cannot resist to give the "experimentalist" point of view, which is often hard to find in the CCL. Best regards, Jose Ignacio -- ******************************************************************************* Dr. Jose Ignacio Garcia-Laureiro Phone : 34-(9)76-350475 Departamento de Quimica Organica Fax : 34-(9)76-567920 Instituto de Ciencia de Materiales de Aragon e-mail: jig@qorg.unizar.es C.S.I.C.-Universidad de Zaragoza jig@msf.unizar.es E-50009 ZARAGOZA (SPAIN) ******************************************************************************* "And all this science I don't understand it's just my job five days a week..." ELTON JOHN - Rocket man ******************************************************************************* >From SERGEI@ps1515.chemie.uni-marburg.de Thu Feb 16 10:36:54 1995 Dear Miquel, Although I do not know exactly the reason for such behaviour of the electron density Laplacian in CO molecule, I have some comments to this subject: 1. Positive values of the Laplacian are typical of C-O multiple bonds as well as of C-F and some other bonds which are normally treated as covalent. The most famous example is F2 where the bond is defi- nitely purely covalent. Probably the positive Laplacian can occur in "electron-rich" molecules. 2. The Bader's topological theory does not state that the bonds with the positive Laplacian are ionic ones. Bader distinguishes between covalent and "closed-shell" interactions. The latter include ionic, donor-acceptor, and van der Waals interactions. 3. It is quite incorrect to compare the absolute values of the Lapla- cian in the bonds formed by different atoms (one cannot say whether LiF is more or less ionic than CO) 4. There exists another criterion of covalency suggested by Cremer and Kraka. Namely, a bond is to be considered as covalent, if the elec- tronic energy density is negative. In CO molecule this is the case. I would not like to discuss whether or not the Cremer-Kraka's cri- terion is better than Bader's one. 5. I have repeated you calculation (QCISD/6-311++G**) of CO molecule at C-O separation of 1 angstroem using Gaussian92 followed by Ba- der's EXTREME program and obtained the value of the electron den- sity Laplacian at the (3,-1) critical point as much as +2.837au. Thus, either in my or in you calculation something is wrong. Yours sincerely, Sergei Vyboishchikov FB Chemie der Universitaet Marburg Hans-Meerwein-Strasse D-35032 Marburg, Germany E-Mail: sergei@ps1515.chemie.uni-marburg.de >From D.A.Buttar@bath.ac.uk Thu Feb 16 11:12:29 1995 Miquel, If i understand the Bader Analysis correctly a positive value of the Laplacian does indicate ionic character. However the magnitude of the charge density at the bond critical point must also be considered. For LiF i would expect that the charge density is much smaller at the bcp than in the corresponding CO bond and that this indicates an ionic LiF bond whereas the positive value for the CO indicates a strongly polar, covalent bond. Classification of bonds may also be performed using the energy density at the bond critical point, as discussed by Cremer and Gauss, JACS, 107, 1985, p.3800. I would be very interested in any other replies you receive on the subject. Cheers, David. >From csonka@iris.inc.bme.hu Thu Feb 16 11:49:02 1995 FTP Hello Miquel, I suppose that you should analyse larger region around the BCP. The analysis of gradient trajectories would help. I think you will find strange behaviour inside the CO electron density near to the critical point. Try acetilene too. Gabor Gabor I. Csonka Budapest University of Technology Tel/FAX: (361) 463.18.35 Inorganic Chemistry Dept. Ch. Bldg csonka@iris.inc.bme.hu H-1111, Bp. Szent Gellert ter 4 http://www.fsz.bme.hu/bme/chemical/csonka.html >From dimas@hobbit.quimica.uniovi.es Thu Feb 16 15:26:08 1995 Hola Miquel, Creo que el libro de Bader (Atoms in Molecules. A Quantum Theory) puede serte util para aclarar tu cuestion. Si mal no recuerdo, en el libro se seganala que las propiedades de la densidad elec- tronica en los puntos criticos de enlace, no son directamente comparables para enlaces constiudos por diferentes pares de atomos. Asi, la magnitud de la densidad electronica de enlace para un C-S no puede comparase con el valor correspondiente en un enlace C-C por ejemplo, para poder decir "quien es mas fuerte que quien". (La teoria no puede respaldar ese tipo de afirmaciones, aunque luego el mismo Bader sin ir mas lejos compare distintos enlaces de puente de H en base al valor de rho(critico) de un modo "empirico"). Ademas, la circunstancia de que el enlace C-O, C-S, presente valores positivos de la Laplaciana, del tipo de los compuestos ionicos LiF, tambien es tratada por Bader. Estos enlaces, C-O, son covalente-polares, implicando una parcial transferencia de carga entre los atomos implicados, de resultas del cual la energia cinetica predomina en la region de enlace. De todos modos, que se trata de una interaccion covalente tambien se puede poner de manifiesto mediante un mapa de la Laplaciana sobre la region de enlace. En dicho mapa se observa que la densidad electronica de C y O se deforma claramente en la region de enlace, mientas que para el caso de compuestos ionicos, lo que se observa es que la laplaciana es practicamente esferica ("dos bolitas") en torno a los atomos en cuestion. Supongo tambien, que si has calculado las cargas atomicas de Bader para el caso del CO y el LiF, tambien tendras otro criterio para distinguir ambos casos. Supongo que en el libro de Bader (cap 7, si mal no recuerdo) podras encontrar detalladamente este tipo de discusion, acerca de la caracterizacion de interacciones atomicas de distintos tipos a partir de las propiedades de los puntos criticos de enlace. Asimismo, la densidad de energia en el punto critico puede ser otra magnitud que diferencie ambos enlaces.(Esta tecnica ya no es de Bader, sino de D. Cremer y E. Kraka y esta expuesta en un par de articulos de agno 1984, en JACS y Angewandte..) Bueno, no se si estos comentarios te pueden servir de algo,... Hasta luego,... Dimas Suarez e-mail: dimas@hobbit.quimica.uniovi.es >From !rgab@netcom.com Thu Feb 16 17:47:22 1995 Reply to: RE>CCL:Bader analysis of CO Miquel: You should read the section(s) in Richard Bader's book (at least) on this matter. I do not believe that there is any correlation, per se, between the mag- nitude of the Laplacian and any chemical notion of 'covalency/ionicity'. The Laplacian is simply the trace of the Hessian of Rho; at the bond- critical point, you look at it as the sum of the 3 eigenvalues (of that Hessian) and conclude: is the sum of the transverse curvatures greater in magnitude than the longitudinal curvature? You should note, as follows, there are two 'extremes' of behaviour of the charge density in 'interatomic' regions: "shared" interactions (typically between covalently bound atoms) in which the Laplacian is _usually_ negative; "closed shell" interactions (like noble-gas vdW dimers and 'ionic' systems) in which it is _invariably_ positive. In between there are so-called "intermediate" cases, like CO (in fact most C=O and C=S bonds) in which the interaction is "shared" but the Laplacian is +ve. You have to also look at the magnitude of Rho itself which is usually around 10^-1 a.u. for "shared" interactions. The reason for the sign of the Laplacian in these intermediate cases is that the location of the 'bond critical point' is close to the "VSCC" (Valence shell charge concentration) of one of the atoms - see Bader's book. The point is that you have to look at the _whole_ picture; in the Theory of Atoms in Molecules, it is often not useful to draw drastic conclusions from single numbers. Many people are guilty of that and misinter- pretations are legion. Another paper you should look at is Bader and Essen, J. Chem. Phys., 1984. I also have to ask you: although your numbers look OK to me, you should check that those values of Del^2 do in fact correspond to the sum of eigenvalues, if you are using the program "SADD2R" (or "EXTD2R"). E-mail me if you think this is a problem. Richard Bone __________________________________________________________ Richard G. A. Bone, PhD. Computational Chemist Terrapin Technologies, Inc. South San Francisco USA E-mail rgab@trpntech.com >From laidig@erebus.mchem.washington.edu Thu Feb 16 18:27:30 1995 Greetings Miquel: You asked about link between the properties of (3,-1) critical points and the character of bonding; specifically about the sign of the Laplacian and the 'ionic character' of bonding in LiF and CO. While its true that the character of chemical bonding can be summarized by the properties at the 'bond critical point' (BCP), I think its unwise to treat the values as magic numbers which tells all about the nature of interaction between two atoms. One must take stock of the entire set of data to get the correct results from these characterizing properties. The rule of thumb is that a low amount of charge AND a positive Laplacian signify an ionic interaction. In the case of LiF vs. CO, the magnitude of the Laplacian at the BCP of CO is more positive, but there is a good deal more electronic charge than typically found in 'ionic bonds' (see Bader and Essen, JCP, 1984). LiF has the small amount of charge typically found in the interatomic region of ionic interactions. LITHIUM FLUORIDE HF/6-311++G** CARBON MONOXIDE HF/6-311++G** RHO 7.0078622807E-02 RHO 5.1842212832E-01 DEL2 7.2110756059E-01 DEL2 1.0834101333E+00 So, the bonding in CO is not the same as that in LiF, 'ionic'. ---Further ramblings--- In fact, the case of CO is particularly interesting as the Laplacian makes clear the seeming contradiction of the different electronegativities of C & O and the small dipole moment for this system. The small dipole moment of CO would lead you to believe that the atoms were roughly electroneutral, based upon the simple idea that the dipole moments is the result of separation of charge. But then how does one reconcile this with the large expected electronegatvities of the two atoms? There is a large transfer of charge from C --> O. But there is also a large opposing polarization of the remaining charge on the carbon atom (unlike what one would expect for spherical ions). It is this large opposing polarization which produces a large dampening of the dipole moments, canceling the contribution from charge transfer, and making the overall dipole nearly zero. This can be seem in the Laplacian map of CO in which the valence shell of carbon is polarized away from the direction of charge transfer, producing a large, relatively diffuse region of charge concentration (see the cover of Bader's paperback to see the Laplacian of CO (blue)). This is why the carbon of CO is so reactive. In addition, the polarization of the carbon valence shell away from oxygen is the reason that the Laplacian is positive at the bond critical point. The Laplacian map of CO shows that the valence shell of O reaches almost to the interatomic surface and that the valence shell of carbon is polarized away from the interatomic surface. So, it is this polarization of C away from O that gives rise to the relatively large value of rho at BCP (a 'sort-of' shared interaction) and the larger, positive magnitude of the Laplacian. Anyway, I think its all pretty interesting... Keith ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Keith E. Laidig Department of Medicinal Chemistry, BG-20, University of Washington, Seattle, WA 98195 E-Mail: laidig@erebus.mchem.washington.edu Phone: (206) 616-2780 or 685-1000 FAX #: (206) 685-3252 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >From ehlers@chem.vu.nl Thu Feb 16 09:28:26 1995 Dear Miquel, instaed of discussing the Laplacian, it is better to compare the energy-density at the bond kritical point (This is the negative K-Value in the Bader-output). For references see: D. Cremer, E. Kraka, Angew. Chem. 8, 614 (1984) (german edition) G. Frenking, D. Cremer, Stucture and Bonding 73, Springer Verlag, Berlin-Heidelberg (1990) W. Koch, G. Frenking, J. Gauss, D. Cremer, J. R. Collins, J. Am. Chem. Soc. 109 5919 (1987). I hope this helps, with best regards, Andreas Ehlers =========================================================================== = Andreas Ehlers - Afdeling Theoretische Chemie, Faculteit Scheikunde = = - Vrije Universiteit Amsterdam = = - De Boelelaan 1083, 1081 HV Amsterdam = = - [ ehlers@iodine.chem.vu.nl = =========================================================================== From pfriedma@acnet.pratt.edu Fri Feb 17 23:50:12 1995 Received: from acnet.pratt.edu for pfriedma@acnet.pratt.edu by www.ccl.net (8.6.9/930601.1506) id XAA03796; Fri, 17 Feb 1995 23:00:17 -0500 Received: by acnet.pratt.edu (5.0/SMI-SVR4) id AA17192; Fri, 17 Feb 1995 22:56:49 -0500 Date: Fri, 17 Feb 1995 22:56:48 -0500 (EST) From: Paul Friedman X-Sender: pfriedma@acnet To: Peter Gedeck Cc: chemistry@ccl.net, Peter Gedeck Subject: Re: CCL:Announce: MolEdit 1.0 In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII content-length: 3826 Sounds very useful. Will try it. Paul Friedman On Fri, 17 Feb 1995, Peter Gedeck wrote: > > Hi, > > I thought it's time to announce the new release of my program MolEdit. This > is an editor for the MS-DOS system, below you will find a description of > its features. > > MolEdit is freeware, further information is available at the WWW page: > http://pctc.chemie.uni-erlangen.de/~gedeck/moledit.html > or download the program from > ftp://pctc.chemie.uni-erlangen.de/pub/moled > > If you download and use it, send me a note, so that I can keep you > informed about bug-fixes. Also, please read the copyright in the > readme-file. > > Peter > > > > --------------------------------------------------------------------------- > > > MolEdit 1.0 > *********** > > Editor for Mopac-Files > ====================== > > > Introduction > ============ > > MolEdit is an editor specially designed to prepare input-files for > quantum-chemical programs. During the construction of an input-file it is > always possible to check the correctness of the defined structure. In case > of errors MolEdit provides informative error messages for the implemented > input formats. > > In addition the combination of the editor with a molecule-viewer allows to > examine the defined geometrical structure. This is especially valuable for > the creation of z-matrices. The molecule-viewer allows to rotate and to > translate the molecule. The structure can be displayed in five different > ways: > > o Wire-frame > o Ball-and-Stick > o Dot-clouds > o Grid > o Spheres > > The combination of an editor with a molecule-viewer will make it very easy > for beginning students to learn to write z-matrices. > > Other features included in MolEdit are: > > o Formatting input-files > o Converting between different formats (Mopac <=> GIP) > o Deleting Dummy-Atoms > o Inserting and deleting of atoms (while keeping the references correct). > > In the current version MolEdit fully supports several file formats itself. > These are: > > o The Mopac file format, which is used by the semiempirical program packages > Ampac, Mopac and Vamp. > o The GIP (geometry-interchange-program) format > o The Xmol format > > Other file formats, among those the Gaussian-file formats are accessible > through the conversion utility Babel. MolEdit provides a interface that > simplifies the usage of Babel. > > MolEdit is also able to load, edit and save UNIX-Files. The is useful, if > UNIX-filesystems are mounted via NFS. > > MolEdit is a DOS-application written for the IBM-PC-AT and needs a VGA or > EGA-card, either monochrome or color. A coprocessor is recommended but not > necessary. In case the real-mode-version of the program is used, either > the use of expanded-memory or of a disk-cache is recommended. The program > was tested with monochrome-VGA-cards and should adjust itself to this > mode. > > ------------------------------------------------------------------------- > > Peter Gedeck > Inst. f. Physikalische Chemie I > Egerlandstrasse 3 > 91058 Erlangen > Germany > > Tel: ++9131 - 85 7335 Fax: ++9131 - 85 8307 > E-Mail: gedeck@pctc.chemie.uni-erlangen.de > WWW: http://pctc.chemie.uni-erlangen.de/~gedeck/gedeck.html > > > -------This is added Automatically by the Software-------- > -- Original Sender Envelope Address: gedeck@pctc.chemie.uni-erlangen.de > -- Original Sender From: Address: gedeck@pctc.chemie.uni-erlangen.de > CHEMISTRY@ccl.net -- everyone | CHEMISTRY-REQUEST@ccl.net -- coordinator > MAILSERV@ccl.net: HELP CHEMISTRY | Gopher: www.ccl.net 73 > Anon. ftp www.ccl.net | CHEMISTRY-SEARCH@ccl.net -- archive search > http://www.ccl.net/chemistry.html | for info send: HELP SEARCH to MAILSERV > >