From DENNYBJ@KIRK.VAX.ASTON.AC.UK Fri Jan 8 09:31:32 1993 Message-Id: <199301081350.AA06002@oscsunb.ccl.net> Date: Fri, 8 Jan 93 13:50 BST From: DENNYBJ@kirk.vax.aston.ac.uk To: CHEMISTRY Subject: force constants from GAMESS & MOPAC Jan. 8, 1993 Netters Does anybody know how to calculate Force Constants from MOPAC or GAMESS outputs. I checked in the MOPAC-6 manual, they gave an example of a FORCE calculation on Formaldehyde with the resulting force matrix for the 4 atoms. How do you go from this to the actual force constants for bonds, angles, torsions, impropers etc. Does anyone know the conversion factors to change from Millidynes/Angstrom or Hartree/bohr*bohr*dalton to Millidyne*Ang/Radian*radian, or to kcal/mol*Ang*Ang or kcal/mol*radian*radian. Similarly for GAMESS, how do you get the bond,angle,torsion and improper force constants from the Hessian matrix. I am working on triazene systems R-NH-N=N-CH3; R-N=N-NH-CH3, R-NH-NH-N=CH2 does anybody have any experience or references on these sorts of molecules Many thanks Brian J. Denny >FROM Brian J Denny Pharmaceutical sciences Aston University Birmingham B4 7ET U.K. From oosten@rulhm1.LeidenUniv.nl Fri Jan 8 15:53:06 1993 Date: Fri, 8 Jan 93 14:53:06 +0100 From: oosten@rulhm1.LeidenUniv.nl (A.B. van Oosten) Message-Id: <9301081353.AA07856@rulhm1.LeidenUniv.nl> To: chemistry@ccl.net Subject: z-matrix and geometry optimization dear netters For some time I have wanted to construct a z-matrix for geometry optimization with the GAMESS package. I use the cluster approach to solid state problems. My present problem occurs for the C2v-symmetric substitutional oxygen impurity in silicon - better known as the oxygen-vacancy center. Take for instance the hydrogen terminated cluster OSi4H12, which is taken to be Td symmetric if the O-atom is left out. I want to optimize the position of the O-atom on the z-axis, with the other positions kept fixed. It's rather tricky to write a z-matrix for this task and the ones I produced were not accepted by GAMESS because e.g. they featured 180 degree angles. Also, I could not use the GEOMETRY option, in which case all positions are optimized. So I proceeded to optimize 'by hand'. This approach can no longer be used if there are more than one geometry variables; e.g. for a simultaneous optimization of O and its two Si neighbours. My request : how should I construct a z-matrix for such a task ? From SC18@NEMOMUS.bitnet Fri Jan 8 03:30:00 1993 Message-Id: <08JAN93.09180326.0032.MUSIC@NEMOMUS> Date: Fri, 08 Jan 93 08:30:00 EST From: SC18000 To: Subject: RE:HELP!! Dear Netters, I need your help about a situtaion that threatens our faculty policies at NOrtheast Missouri State University. I will describe the situation and ask for your opinions, information, general direction and hopefully sympathetic ears. We do not have a chemistry department at NMSU, but are in a divisional organization. In this organization one man rules four or five discip- lines. Each discipline is supposed to be free to make their own policys Chemistry has a policy that no student can graduate with less than a C in a required core course. In fact the division head just caused us to put all substitution and waiver policies in writing to prevent abuse by individual faculty for favorite students. I taught a 1 hour Org. II lab that is a required course for all chemistry majors. One student sys tematically refused to do more than 50% of the work. He was a senior a severe attitude problem noted by other faculty. He was to graduate in Dec., but I gave him (quite justifiably) a D. The grade in not in question. All agree that I am fair. The division head was approached this student, and as a result the faculty policy was unilaterally waived by the division head so as to let the student graduate. I'm not mean by nature, but when I asked the division head about it he said simply "I it by myself. I made the decision that a 1 hour course should not make a student not graduate." Will you help referee this situation? I would appreciate all opinion about what I perceive to be a serious breach of ethics, and probable for a charge of fraud. Please help! How much of this sort of thing goes on? How much shuld we tolerate. I am a full professor with 20+ years and have never seen anything so blatant. Sincerely, Ken Fountain sc18@NEMOMUS From JOHNSON_DOUGLAS_W@INET.D48.LILLY.COM Fri Jan 8 04:19:39 1993 Date: 08 Jan 1993 09:19:39 -0500 (EST) From: "dwj@lilly.com (Doug Johnson MC7R7)" Subject: replies to my posting about PDB searching methods To: chemistry@ccl.net Message-Id: <01GT9RKR18Z68WX839@INET.D48.LILLY.COM> The following are what I received in reply to my question about searching the Brookhaven files. I think that the easiest is to use Laura Walsh's annotated list method; any type of pattern searching on her list should work; I've even used Microsoft Word's "Find" facility on it to search it. She is working on an updated list at this time and will post it to OSC when finished. Contact her directly if you want more info. ++++++++++++++ Reply #1: Two scripts were written by Vamsee Lakamsani and they follow; the first is sort of an interactive searching method which may be tedious to use in a directory containing over 900 entries. The second is a similar search which writes the results into a file which you can look at when the search is done. +++++++++++++++ Dr.Johnson, I am computer science Master's student very much interested in computational biochemistry. I saw your note 10 minutes ago and hacked out this script. I am not sure it meets your requirements but it does the following. 1. Prints the HEADER and COMPND of each pdb file. 2. Prompts you to accept or reject this file. You will look at the header field and decide if its a DNA structure or not. 3. If you accept this file then the pdb file name is written into an output file called "find.out". 4. If you would like to make this selection automatic then you should give me more information about what keywords to expect if the structure is a DNA. This script expects pdb file names to be of the format pdb1acb.ent. I tested this on a Silicon Graphics IRIS 4.0.5A and we had April 1992 version of the PDB. Assuming your pdb directory is called ./PDB. Then, %cp script ./PDB %cd ./PDB %chmod +x script %./script If you would like me to make modifications I will be glad to do that. =================== CUT HERE ========================= #!/bin/csh # C shell script to extract pdb headers, Vamsee Lakamsani(vamsee@tamu.edu) if(-e find.out) then /bin/rm find.out endif echo "Results" > find.out echo "==================" >> find.out foreach num ( 1 2 3 4 5 6 7 8 9 ) foreach file (`/bin/ls pdb"$num"*.ent`) sed -ne /HEADER/p $file sed -ne /COMPND/p $file echo "Do you like this file?(y/n)\c" if(`line` == y) then echo $file >> find.out endif echo " " echo " " end end ================== END OF script ================ Vamsee -- Vamsee Lakamsani Center for Macromolecular Design & Dept. of Computer Science Texas A&M University, College Station, TX 77840 Fax: (409) 845 6980 Internet: vamsee@tamu.edu ++++++++++++++++ Reply #2 (also from Vamsee) This one does the search and writes to a file. +++++++++++++++ #!/bin/csh # C shell script to search for patterns in PDB files. # Vamsee Lakamsani(vamsee@tamu.edu). # Usage: pdbfind # Make pdbfind executable. # Copy pdbfind to the directory in which the PDB is located. # Issue the command ./pdbfind # Look at the output file pdbfind.out if($#argv != 1) then echo "Usage: pdbfind " exit 1 endif if(-e pdbfind.out) then /bin/rm pdbfind.out endif echo `date` > pdbfind.out echo "The following structures had the pattern $1 " >> pdbfind.out echo "==================" >> pdbfind.out # This double loop is necessary because of argument length # limitations in C shell. foreach num ( 1 2 3 4 5 6 7 8 9 ) foreach file (`/bin/ls pdb"$num"*.ent`) set a=`grep $1 $file | sed -e s/B\-//` if($#a) then echo $file >> pdbfind.out echo "Found Keyword DNA in $file." endif end end ++++++++++++++++++ Reply #3 Sort of a custom software solution to this problem: +++++++++++++++++ Dear Douglas, I have seen your recent posting on the network concerning concerning searching the Brookhaven databank for items of interest. We recently announced a molecular graphics package INTERCHEM on the net, and this includes a program PROTEINS which has a search facility. The main function of PROTEINS is in making files suitable for display by INTERCHEM, but there is an index which can be searched manually page by page, and also a keyword search. This works on a file which is constructed by concatenating all the non-numeric data in the databank (roughly everthing except ATOM CONECT, and a few other types). The whole package (INTERCHEM, PIFF [molecular mechanics], PROTEINS, and CONVERT [inter-conversion of various data formats]) is available through QCPE and costs $2000 to commercial organisations. You could write to QCPE and see if they will sell PROTEINS separately. An alternative is for me to send you by ftp the concatenated file (this is about 2.3 Mbytes long in compressed format, expanding up to about 7 Mbytes when uncompressed). This should be searchable using "grep". The only problem is that the index and the concatenated file need to be re-made every time Brookhaven updates their files (or when you acquire an updated version). Programs for doing this are included with the complete package. If you want the concatenated file please send me your ethernet address IN NUMERIC FORM (ie www.xxx.yy.zz). If you want details of the package I can send them, but they are also deposited in the archives at ccl.net. Yours sincerely Peter Bladon Return-path: ++++++++++++++++++ Reply #4 >From Laura Walsh at Beckman institute, at UIUC, her annotated list method. I have an annotated PDB listing that may be of use to you. There is an older version available from the Computational Chemistry list, but I am working on a newer version which includes the latest release of the Brookhaven files. Let me know if you would like a copy. Laura Walsh Laura Lynn Walsh (lwalsh@nemo.life.uiuc.edu) (217) 244-2894 Beckman Inst., Univ. of IL * 405 N. Mathews Ave. * Urbana, IL 61801, USA ++++++++++++++++++ Reply #5 I couldn't get this to work; I got an "Arguments too long" message from the system. I think there are too many files in the directory for this to work properly, but in principle, I think it should work OK. ++++++++++++++++++ To search all pdb files for lines beginning with HEADER or COMPND is a simple UNIX oneliner. If you don't have egrep you can use grep instead, but egrep is faster. Both egrep and grep are case sensitive. egrep '^HEADER|^COMPND' *.pdb search for a line that begins with HEADER or begins with COMPND in all files with a .pdb extension If you are only interested in pdb files with HEADER or COMPND lines that contains a word (say DNA) then you can pipe the result of the line into a new search: egrep '^HEADER|^COMPND' *.pdb | egrep 'DNA|dna' Good Luck! -- Anders Sundin e-mail: Anders.Sundin@orgk2.lth.se Organic Chemistry 2 sundinKC@dna.lth.se Lund University ok2aps@seldc52.bitnet P.O. Box 124 voice: +46 46 104130 S-22100 Lund, Sweden fax: +46 46 108209 ++++++++++++++++++++++++ These are all the replies I received as of Friday morning Jan. 8, 1993. Hope they are of some help to you. Douglas W. Johnson Lilly Research Labs Indianapolis, IN 46285 dwj@lilly.com From: JOHNSON DOUGLAS W (MCVAX0::RX23376) To: VMS MAIL ADDRESSEE (IN::"chemistry@ccl.net") From topper@haydn.chm.uri.edu Fri Jan 8 06:22:47 1993 Date: Fri, 8 Jan 93 11:22:47 -0500 From: topper@haydn.chm.uri.edu (Robert Q. Topper) Message-Id: <9301081622.AA17905@haydn.chm.uri.edu> To: DENNYBJ@kirk.vax.aston.ac.uk, CHEMISTRY@ccl.net Subject: Re: force constants from GAMESS & MOPAC With respect to Brian Denny's question about force constants: First, I should mention that I have never run MOPAC or GAMESS myself, and so I may be mistaken about the force constants they produce. However, I am pretty familiar with some of the conventions which are in the literature for force constants. I would like to point out that the problem is more complex than simply "converting" from one set of units to another; a coordinate transformation between "internal" and "normal" coordinates is also required. First, there is no such thing as an "actual" force constant. However, one way to define force constants is to expand the potential in terms of some arbitrarily chosen set of internal coordinates (which, by the way, are not unique), and then call the coefficients of such an expansion "internal force constants." Another, equally correct, way to expand the potential is to use a set of "normal" coordinates, as described in a number of texts (see Wilson, Decius and Cross' classic text "Molecular Vibrations" from the fine folks at Dover Press, or G. Herzberg's text on IR and Raman Spectroscopy). This is the conventional expansion used by vibrational spectroscopists, because the force constants obtained in such an expansion are directly related to the harmonic frequencies of the molecule, and thus help one assign an IR or Raman spectrum. However, those of us doing simulations generally prefer to use internal coordinate systems to construct potential energy surfaces, and so one needs to transform coordinates to accomplish this. So, Brian has 2 problems: (1) the relationship between conventional "normal" coordinates and his favorite set of "internal" coordinates, and (2) converting between atomic units (Hartree, Bohr) and his favorite units of energy and length (Kcal/mole, Angstroms). The second is easy; check the August 1992 issue of PHYSICS TODAY for conversion factors from atomic units (Hartree, Bohr) to "normal" units (Joules, Angstroms). Also, look at Szabo & Ostlund's book "Modern Quantum Chemistry: Introduction to Advanced Electronic Structure Theory" on page 42 for a definition of atomic units, which are used in virtually all electronic structure codes (and I think all other codes should use them too! But that's an editorial comment). Problem (1) requires a knowledge of the mathematical relationship between a set of internal coordinates and a set of normal coordinates. A hint: if you know the relationship between cartesians and normals, and the relationship between cartesians and internals, you can get the relationship between internals and normals. But you have to keep careful track of the rotations and translations as well when you work this way, and some matrix algebra is involved. So you may not want to do this... Another, much easier way is to just look up the relations as given in Wilson, Decius and Cross, and in Cotton's book, "Chemical Applications of Group Theory," Second edition. A third possibility (which only a GAMESS or MOPAC maven would be aware of) is that there might exist a keyword in one of these programs which outputs force constants in the form that you want. Finally, unless I am mistaken, SURVIBTM, the fine program by Ermler, Hsieh, and Harding (Comput. Phys. Commun. 51, 257 (1988) can be used to accomplish the transformation (but you might need to give it a grid of ab initio points; I'm not positive). Good luck! If anyone (including Brian) is interested, I can supply a preprint of an article in which the relationship between various coordinate systems and force constants is presented in some detail (incidental to carrying out vibrational perturbation theory calculations on triatomic molecules using SURVIBTM). :-) - Robert ******************************** * Robert Q. Topper * * Department of Chemistry * * University of Rhode Island * * Kingston, RI 02881 * ******************************** * rtopper@chm.uri.edu OR * * topper@haydn.chm.uri.edu * * (401) 792-2597 [office] * * (401) 792-5072 [FAX] * ******************************** From SL1HOLL@CUCHEM.bitnet Fri Jan 8 07:29:00 1993 Message-Id: <199301081954.AA11357@oscsunb.ccl.net> Date: Fri, 8 Jan 93 12:29 EST From: Subject: PC Use Survey To: chemistry@ccl.net We are in the process of getting a PC to run/develop chemistry modeling programs and would like to draw on the experience of the viewers of CCL. A survey follows which is an attempt to determine what type of PC's people have, what they are primarily used for, and what their future PC plans are (important for our development goals). As with most things in life we are pressed for time because we would like to get a PC as soon as possible, so I will be collecting responses until next Friday (1/15/93). If sufficient responses are made I will post a summary to the list. Please send all responses to: sl1holl@cuchem ******************************************************* Your company: Commercial [ ] Academic [ ] PC Company Manufacturer: Type of Processor: Amount of Memory: Amount of Disk: 1. Are you running DOS? Windows? Version? 2. Is your PC networked? If so, what type of network? 3. What are the main software packages that are used? 4. Which molecular modeling or chemistry packages do you use? 5. Do you run any X-Windows emulator packages? If so, which one(s) and what do you think of it (them)? 6. What type of graphics? Mono, CGA, VGA, SVGA, etc.? 7. How satisfied have you been with your machine(s)? 8. Do you plan to buy more PC's in the future? What type(s)? What type of graphics? 9. What are your PC's mainly used for? 10. Other hardware (Mac's, Workstations, numbers and type of each): 11. Other comments: ___________________________________________________________ Frank Hollinger Columbia University email sl1holl@cuchem