From: Klaus-D. Warzecha [warzecha@cityweb.de] Sent: Tuesday, September 08, 1998 5:52 AM To: hyperchem@hyper.com Subject: Reconstruction of HIN files from a SNP ? Dear HyperChemists, is it possible to reconstruct a series of HIN files from a plain SNP file obtained in a MD simulation ? It appears, that the snapshot file neither contains information on the force field used, nor the full connectivity matrix for each atom. (At least I wasn't able to extract the latter so far.) Thus, one seemingly has to use an existing HIN file as a template and just change the coordinates. Yours sincerely, Klaus-D. Warzecha From: Klaus-D. Warzecha [warzecha@cityweb.de] Sent: Sunday, September 20, 1998 3:26 PM To: hyperchem@hyper.com Subject: snapshot files / info on MM or QM calculation Dear HyperChemists, concerning the snapshot files generated in MD simulations I'd like to a ask for some information. Supposed one only has a *.snp files and neither a *.hin with a reference to it (dynamics foo.snp line) nor the log file of that particular run. Is it possible to decide whether the simulation was performed on molecular mechanics (MM) or quantum mechanical (QM) level other that by checking for a change in QM-relevant data ? The 172 byte header block of a *.snp seemingly has no entry for the calculation type (QM/MM), the particular force field (MM+, AMBER, etc.) or QM approach (AM1, PM3, etc.) used. The most straight-forward approach thus appears to be reading of some 3 (first, middle, last) or more trailer blocks in the *.snp and comparison of data which should differ from zero only in the case of a QM run (such as HOF, semiempirical atomic/binding/core energy, etc.). It might however be true that the requested information is already coded within the reserved entries of the header block. Any hint on that topic is highly appreciated. Greetings, Klaus-D. Warzecha From: Klaus-D. Warzecha [warzecha@cityweb.de] Sent: Sunday, September 20, 1998 3:26 PM To: hyperchem@hyper.com Subject: HAPI, use of hbExecBin with parameters Dear HyperChemists, just a little question concerning the use of the HAPI.DLL with VisualBasic. While passing of parameters a hbExecTxt is pretty obvious (plain concatenation of command and parameters), I'm pretty lost with the same in binary mode. According to the docs, such a call would be something like Dim lngRetVal as Long Dim Params as IntBuff Dim Size as Long lngRetVal = hbExecBin (cmd, Params, Size). As long as the data passed simply is an Integer, Real or String, it is indeed easier to use the SetInt, SetReal ... functions instead, but how would one pack these data to this strange IntBuff parameter, or more specific : how would it look like when calling a restraint command in binary mode, where a string and two reals have to be stuffed into the IntBuff ? Greetings, Klaus-D. Warzecha From: Márcio Cyrillo [cyrillo@ifi.unicamp.br] Sent: Monday, September 21, 1998 11:42 AM To: hyperchem@hyper.com Cc: CHEMISTRY@infomeister.osc.edu Subject: Restraining dihedrals using HyperChem Dear HyperChemists, I would like to ask you something that I think is very strange about HyperChem Geometry Optimization (GO) calculations. Actually I had already asked this before but the latter support man from HyperCube (Dr. Stravrev) did not want or did not know how to answer to my question. I think its about time to clear this issue! I hope someone can help me. The problem is: we read in HyperChem manual that for GO optimizations we can restraint a variable (bond length, angle or dihedral) using a restraing constant that adds a force field to the hamiltonian of the system so that a potential like that of a spring can "hold" the variable to a specific supplied value. This "Hook" constant, default value equals 60, is set up depending on the problem, but in the manual they tell us that it cannot be as big as we want because we would then add a nonlinear term that would perturbate the original hamiltonian producing artificial results. I have the following questions: · For molecular mechanics I understand this procedure of adding a quadratic potential to the hamiltonian to restrain a variable, but it seems to me as being meaningless to apply the same procedure to a semi-empirical hamiltonian. Am I wrong? · In mopac we can easily accomplish the restraint of a variable - changing a digit from 1 to 0 in the z-matrix scheme. Why did not HyperChem use the same algorithm? With MOPAC it works fine, which is pretty different in HyperChem where it almost impossible to, in fact, restrain the variable to the desired value. Everyone that has used my program HyperSpin (www.ifi.unicamp.br) before, knows what I mean. Talking about HyperSpin, I am going to release a new version soon but I am also working on a interface that will do quite the same work but using old and good MOPAC6. I am looking forward to hearing from you soon, best regards, Marcio. -- Marcio Cyrillo - http://www.ifi.unicamp.br/~cyrillo email: cyrillo@ifi.unicamp.br or mcyrillo@yahoo.com ICQ: 14059279 Graduate Student State University of Campinas (UNICAMP), SP, Brazil Applied Physics Department - DFA/IFGW room 51, phone +55 19 788 5364, fax +55 19 788 5376 From: Márcio Cyrillo [cyrillo@ifi.unicamp.br] Sent: Tuesday, September 22, 1998 10:09 AM To: hyperchem@hyper.com Subject: HC GOs Dear all, First I would like to thank everyone who has answered to my question concerning HyperChem GOs and restraints. I would like also to apologize for my misunderstandings and/or poor knowledge of some HC capabilities and functionality. With your e-mails I have been able to learn a bit more! Well, what I read about constraints is, in fact, in HC 3.0 manual, I will check it again and quote it later. I couldn't find the same passage in HC 5.0 manual, I do not know why, I'll check it again. I was wrong about the default value of 'restraint force': its 16 and not 60 kcal mol?1degree?2. Thanks to whom has corrected me. But I am sure that if you use a big value for it, lets say 10000, you will be able to restrain a dihedral (for instance), but the optimization will be far more unstable (difficult to converge) than the same one without restraints at all, you can try it for yourselves. Even then, you will find out that the value for dihedral is no xx.xx as you wanted but xx.(xx + dx). Interesting is that the same calculation done with mopac and the dihedral constrained to the same value, will converge fast and smoothly (well at least for the molecules I have tested). I understand now that HC does not work with internal coordinates and that's why it cannot apply the same trick as in mopac (the 0/1). But I still believe that this point must be fully cleared since we must now exactly what is being added to the Hamiltonian and how it can affect our calculations. I will post a summary at the end of this discussion and answer later to all e-mails that I received. Márcio. ps: I do will continue with HyperSpin development and many other programs! -- Marcio Cyrillo - http://www.ifi.unicamp.br/~cyrillo email: cyrillo@ifi.unicamp.br or mcyrillo@yahoo.com ICQ: 14059279 Graduate Student State University of Campinas (UNICAMP), SP, Brazil Applied Physics Department - DFA/IFGW room 51, phone +55 19 788 5364, fax +55 19 788 5376 From: Márcio Cyrillo [cyrillo@ifi.unicamp.br] Sent: Tuesday, September 22, 1998 12:42 PM To: hyperchem@hyper.com Subject: [Fwd: Restraining dihedrals using HyperChem] Hi! I decided to forward this message from Dr. Lazlo that I enjoyed and think will help other HC users. Thanks, MC -- Marcio Cyrillo - http://www.ifi.unicamp.br/~cyrillo email: cyrillo@ifi.unicamp.br or mcyrillo@yahoo.com ICQ: 14059279 Graduate Student State University of Campinas (UNICAMP), SP, Brazil Applied Physics Department - DFA/IFGW room 51, phone +55 19 788 5364, fax +55 19 788 5376 --------------F185251D0087DD8181B948C8 Content-Type: message/rfc822 Content-Disposition: inline Received: (from daemon@localhost) by ifi.unicamp.br (8.8.8/8.8.8) id NAA17931 for ; Tue, 22 Sep 1998 13:22:55 -0300 (BSC) Received: from dub-img-6.compuserve.com(149.174.206.136) via SMTP by lua.ifi.unicamp.br, id smtpdaarhka; Tue Sep 22 13:21:47 1998 Received: (from root@localhost) by dub-img-6.compuserve.com (8.8.6/8.8.6/2.14) id MAA03657 for cyrillo@ifi.unicamp.br; Tue, 22 Sep 1998 12:21:26 -0400 (EDT) Date: Tue, 22 Sep 1998 12:20:55 -0400 From: "CYCLOLAB R&D Lab." Subject: re: Restraining dihedrals using HyperChem Sender: "CYCLOLAB R&D Lab." To: "Márcio Cyrillo" Message-ID: <199809221221_MC2-5A3D-7C12@compuserve.com> MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Disposition: inline X-MIME-Autoconverted: from quoted-printable to 8bit by ifi.unicamp.br id NAA17931 Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by ifi.unicamp.br id QAA24702 Hi M=E1rcio, I have copied here the hlp message of HC about the constraints. I = suppose that you might have misunderstood something. ********** Force constant This constant regulates the strength of the constraint. Larger values = give a stronger constraint. The Default values are appropriate for molecular dynamics calculations. Much larger values (for example, 105) are appropriate for geometry optimization, but these values might require = more optimization cycles. Default The defaults are Length, 7 kcal mol=961=C5=962; Angle, 125 kcal = mol=961 degree=962; and Torsion, 16 kcal mol=961degree=962. Other Enter a force constant. ************ I suppose there is no any words about the force fields, this means = only that during the optimizations steps HyperChem divide (weighted of = course) the atomic movements according to the results of calculatation = (gradient + method). This force constant does not mean the Hook's force constants, means only the weight of the appropriate component in the movements. = I.e. without restraints if the atomic movement in an optimization step is divided equally, all atom has the same weights and all atom is moved accordingly. If we are using a constraint, this means that during an optimization step the restrained distances, angles, dihedrals is being tried to be kept as close as possible to the desired value, while other atoms' movements are adjusted to those small movements. This may result failed optimizations, as well, because there might be such arrangements = of the atoms that the gradient cannot reach the desired value. This work = of the restraints is tru not only for the MM but for SE and ab initio, as well, but in the latter two cases the use of restraints require = different (programmer's) solution, due to the different optimization procedures. I think, this is a very well written part of HyperChem, and it is very convenient to use, if you are intending to keep constant only several parameters. The problems starts when you want to keep planar a benzene ring, etc., because this procedure requires more than one parameter to = be fixed (in case of Bz-ring, not only one dihedral must consrtrained to = zero, at least 3 consecutive torsionals must be constrained -this is a = tiring process in case eg. a protein which contains too much aromatic = sidechain). But this is not so common problem. I hope that now everything is more clear. But, I do not know why = Krassimir has not answered it, because it is trivial. Regards, Laszlo --------------F185251D0087DD8181B948C8-- From: ganain [ganain@pc-hb.uni-sb.de] Sent: Wednesday, September 23, 1998 3:20 PM To: hyperchem@hyper.com Subject: MMOK Hello! It seems that Hyperchem lacks the possibility to constrain the peptide linkage by default. I wonder, if there exists something similar to the MMOK keyword as used by MOPAC. It is somewhat annoying to have to constrain all peptide linkages by oneself, so it would be helpful, if something like the MMOK option could be included in a later version of Hyperchem. Regards, D. Neumann -- Though this be madness, yet there is method in 't. From: Laurence Lavelle [lavelle@mbi.ucla.edu] Sent: Thursday, September 24, 1998 2:45 PM To: ganain; hyperchem@hyper.com Subject: Re: MMOK This type of feature is definitely needed for constraints in general. Especially for polymers like DNA and RNA where a domain needs restraints and the rest is "free" during MD. Currently the number of restraints that the user must specify makes this type of MD not feasible in HC. Laurence At 12:19 AM 9/24/98 +0200, ganain wrote: >Hello! >It seems that Hyperchem lacks the possibility to constrain the peptide >linkage by default. I wonder, if there exists something similar to the >MMOK keyword as used by MOPAC. >It is somewhat annoying to have to constrain all peptide linkages by >oneself, so it would be helpful, if something like the MMOK option could >be included in a later version of Hyperchem. >Regards, >D. Neumann >-- >Though this be madness, yet there is method in 't. > From: Joel Polowin [JPolowin@idsdetection.com] Sent: Friday, September 25, 1998 11:30 AM To: hyperchem@hyper.com Subject: RE: MMOK Laurence Lavelle [SMTP:lavelle@mbi.ucla.edu] wrote > This type of feature is definitely needed for constraints in general. > Especially for polymers like DNA and RNA where a domain needs restraints > and the rest is "free" during MD. Currently the number of restraints > that the user must specify makes this type of MD not feasible in HC. I guess I don't understand what you're trying to do. Can't you just select the atoms that you want to move during MD, and unselect the atoms that you want to remain fixed, and run the MD simulation? Joel jpolowin@idsdetection.com "Pencils and pens and implements of depiction..." From: Joel Polowin [JPolowin@idsdetection.com] Sent: Monday, September 28, 1998 6:23 AM To: hyperchem@hyper.com; Laurence Lavelle Subject: RE: More on MMOK and unselect in HC Laurence Lavelle wrote: > Visualize a piece of DNA (or any polymer). Now you want MD on the complete > molecule and at the same time keep the conformation of some small part > fixed. If you unselect this small part then this part does not move at ALL > and the MD of the rest of the polymer is perturbed by this frozen part. So -- if I'm understanding correctly -- you want to specify a group of atoms that whose positions with respect to each other are fixed, but which can move as a group relative to the other atoms? If you have only one such group, it seems to me that freezing those atoms absolutely but allowing all other atoms to move wouldn't cause much more perturbation than allowing those atoms to move, but only as a block. If you think of that group of atoms as being like a single "super-atom" with a very large effective mass, the motions of that high-mass atom, scaled by the MD calculations, would be so much smaller than the motions of the other atoms that they might almost as well be zero. Joel jpolowin@idsdetection.com From: Laurence Lavelle [lavelle@mbi.ucla.edu] Sent: Friday, September 25, 1998 3:27 PM To: Joel Polowin; hyperchem@hyper.com Subject: More on MMOK and unselect in HC Hi Joel, No, because the unselected atoms are "frozen". Visualize a piece of DNA (or any polymer). Now you want MD on the complete molecule and at the same time keep the conformation of some small part fixed. If you unselect this small part then this part does not move at ALL and the MD of the rest of the polymer is perturbed by this frozen part. What HyperChem should have is the ability to maintain a local conformation but not freeze it. The following two examples should make clear the impact of freezing domains. Take a piece of string and place pins at those positions that you want fixed (in HC they are frozen). Then jiggle the string to see what conformations are possible, and you will get a very small subset of conformations due to the frozen domains (i.e., pin fixing the string to board). The second example is protein folding. Lets say you want the peptide bonds maintained in a fixed geometry. If one uses "unselect in HC" to maintain this geometry, the polypeptide really w'ont do anything because each peptide bond is in fact frozen. The polypeptide can not move. Laurence At 02:30 PM 9/25/98 -0400, you wrote: >Laurence Lavelle [SMTP:lavelle@mbi.ucla.edu] wrote > >> This type of feature is definitely needed for constraints in general. >> Especially for polymers like DNA and RNA where a domain needs >restraints >> and the rest is "free" during MD. Currently the number of restraints >> that the user must specify makes this type of MD not feasible in HC. > >I guess I don't understand what you're trying to do. Can't you just >select the atoms that you want to move during MD, and unselect the >atoms that you want to remain fixed, and run the MD simulation? > >Joel >jpolowin@idsdetection.com >"Pencils and pens and implements of depiction..." > From: Laurence Lavelle [lavelle@mbi.ucla.edu] Sent: Monday, September 28, 1998 4:10 PM To: Joel Polowin; hyperchem@hyper.com Subject: RE: More on MMOK and unselect in HC At 09:22 AM 9/28/98 -0400, Joel Polowin wrote: >Laurence Lavelle wrote: > >> Visualize a piece of DNA (or any polymer). Now you want MD on the >complete >> molecule and at the same time keep the conformation of some small part >> fixed. If you unselect this small part then this part does not move at >ALL >> and the MD of the rest of the polymer is perturbed by this frozen >part. > >So -- if I'm understanding correctly -- you want to specify a group >of atoms that whose positions with respect to each other are fixed, >but which can move as a group relative to the other atoms? Yes. >If you have only one such group, it seems to me that freezing those >atoms absolutely but allowing all other atoms to move wouldn't cause >much more perturbation than allowing those atoms to move, but only >as a block. If you think of that group of atoms as being like a >single "super-atom" with a very large effective mass, the motions >of that high-mass atom, scaled by the MD calculations, would be so >much smaller than the motions of the other atoms that they might >almost as well be zero. No not as a super-atom, but as a group of atoms with restraints that maintain that domains local conformation. For example duplex DNA with 10 base pairs of which 2 base pairs are maintained in the B-conformation while the rest is not restrained during MD. Currently the number of user defined restraints, even for this very simple (or small) example, makes it not feasible in HyperChem 5. Laurence From: Joel Polowin [JPolowin@idsdetection.com] Sent: Tuesday, September 29, 1998 5:40 AM To: hyperchem@hyper.com; Laurence Lavelle Subject: RE: More on MMOK and unselect in HC Laurence Lavelle wrote: > No not as a super-atom, but as a group of atoms with restraints that > maintain that domains local conformation. For example duplex DNA with 10 > base pairs of which 2 base pairs are maintained in the B-conformation > while the rest is not restrained during MD. Currently the number of user > defined restraints, even for this very simple (or small) example, makes > it not feasible in HyperChem 5. What I'm suggesting, though, is that if you *did* apply all of those constraints -- or had it done automatically, or had the same effect added through another software feature -- then the designated atoms would be nearly immobile anyways. If the designated atoms were moved one at a time by the optimizer, than only very tiny motions would be possible because of all of the restraining forces. If the atoms were moved as a group, there would be proportionally more restraining forces. As far as the thermal motions are concerned, with all of the atoms forced to move together, again the motions would have to be very small -- much smaller than the motions of the other atoms, at any rate -- or else the MD model would be unbalanced. (I suppose that rotation of the group of atoms, rather than translation, would be a different matter.) I don't think it's that much more of a perturbation of the MD model to have those atoms fixed in space than to require them to move in concert. Joel jpolowin@idsdetection.com From: Joel Polowin [JPolowin@idsdetection.com] Sent: Tuesday, September 29, 1998 5:28 AM To: hyperchem@hyper.com Subject: RE: HAPI, use of hbExecBin with parameters Klaus-D. Warzecha wrote: > lngRetVal = hbExecBin (cmd, Params, Size). Note that the "cmd" values are mostly not consistent between HyperChem versions 5.0x and 5.1 . A HAPI-based program which uses hbExecBin, or other Bin commands/requests, and has these values compiled into it, won't work for both versions of HyperChem unless the program determines which version of HyperChem it's attached to, and uses the correct command values. Joel jpolowin@idsdetection.com From: Vernon Anderson [anderson@biochemistry.BIOC.cwru.edu] Sent: Tuesday, September 29, 1998 10:09 AM To: Hyperchem@hyper.com Subject: More on MMOK There has been some discussion over using HC and domain motions. I have been interested in using moecular dynamics to follow the motions about a glycine hinge region within a long alpha helix. Because there are two helical regions whose motion I am following relative to each other, simply selecting the hinge region does not permit the domain motion. I have unselected the amino terminal portion of the ackbone up to the hinge, and then have had to use named selections to select the phi and psi dihedral for the carboxy terminus. I have used a large constraint for all of these dihedrals. This process has been aided greatly by the use of a script and the ability of hin files to store selections. The molecular dynamics was "successful" to the extent that a significant range of motion about the gly-gly hinge was observed while the constrained backbone of the alpha-helix remained alpha-helical. My preference in this matter would be to enable the molecular dynamics to be specified in internal cooridnates rather than in Cartesian coordinates as an option and then permitting the internal coordinates to be selected or deselected. Vernon Anderson anderson@biochemistry.cwru.edu Ph: 216-368-2599 fax: 216-368-3419 Case Western Reserve University 10900 Euclid Ave., Cleveland, OH 44106-4935