From owner-chemistry@ccl.net Sat Jun 6 05:28:00 2009 From: "Michel Petitjean petitjean.chiral]|[gmail.com" To: CCL Subject: CCL:G: detect permutation of atoms in xyz files Message-Id: <-39455-090606052505-26393-cvaocfg4VKboJcBhkRAXNA_._server.ccl.net> X-Original-From: Michel Petitjean Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1 Date: Sat, 6 Jun 2009 11:24:53 +0200 MIME-Version: 1.0 Sent to CCL by: Michel Petitjean [petitjean.chiral-,-gmail.com] Kabsch method is based on the SVD and returns an orthogonal transform, not necessarily a rotation. E.g., if you input two enantiomers, you will get a null minimized RMSD. Quaternion methods are devised: see the appendix in J. Math. Phys. 1999, 40(9), 4587-4595 and the generalization to continuous sets (infinite number of points) and distributions in the appendix A.5 in J. Math. Phys. 2002, 43(8), 4147-4157 (the SVD is in appendix A.2). Michel Petitjean, DSV/iBiTec-S/SB2SM (CNRS URA 2096), CEA Saclay, bat. 528, 91191 Gif-sur-Yvette Cedex, France. Phone: +331 6908 4006 / Fax: +331 6908 4007 E-mail: michel.petitjean%%cea.fr, petitjean.chiral%%gmail.com http://petitjeanmichel.free.fr/itoweb.petitjean.shape.html ---------- Forwarded message ---------- > From: Jerome Kieffer jerome.Kieffer*_*terre-adelie.org Date: 2009/6/5 Subject: CCL:G: detect permutation of atoms in xyz files To: "Petitjean, Michel " Sent to CCL by: Jerome Kieffer [jerome.Kieffer/a\terre-adelie.org] On Fri, 5 Jun 2009 13:51:20 +0300 (EEST) "Berger Raphael berger^-^chem.helsinki.fi" wrote: > My idea would be to first find a standard orientation for both structures > (maybe like Gaussian does; is there a separae tool for that?) and label > the atom names uniquly (Si -> Si1, C -> C3, ...) and sort both of them > lexicographically according to the size of the cartesian coordiantes, > In this way one gets one Permutation functions for each file, which can b= e > combined (and inverted) to get the permutation. I tried the same except that I sorted vs the distance to the center of the molecule, atom type by atom type, then I did a Kabsch-optimised rotation. It works pretty well. Hope this helps, I can even provide the piece of code doing it (but integrated in a bigger software). -- J=E9r=F4me KIEFFER http://www.terre-adelie.org From owner-chemistry@ccl.net Sat Jun 6 06:55:01 2009 From: "Berger Raphael berger,+,chem.helsinki.fi" To: CCL Subject: CCL: detect permutation of atoms in xyz files Message-Id: <-39456-090606065334-26608-995VNH2IwKtpkFCtZ9G9Zw],[server.ccl.net> X-Original-From: Berger Raphael Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Date: Sat, 6 Jun 2009 13:53:02 +0300 (EEST) MIME-Version: 1.0 Sent to CCL by: Berger Raphael [berger^chem.helsinki.fi] Dear Michel, J'er^ome Tom (and CCL-Readers), thank Yo very much for the quick reply and very constructive suggestions! ~~Michel Using CSR is most convienient, since its free available and it works obviously straight ahead. I wrote a small script doing the taking with the CSR programm depending on command line arguments and writing the output in the format i need. Maybe 2 remarks: At the moment its use is quite restricted for me since it requires the old libg2f.so.2.0 library from the g77 site which is seemingly replaced in most new linux distributions by gfortran. It was a bit confusing for me how to understand the sentence in the doc: "to concatenate the two molecules" Thank god my second try just to concatenate two babel generated .mol2 files worked out well (while the first try to do same for pdb files was not successfull). ~~Jerome "I tried the same except that I sorted vs the distance to the center of the molecule, atom type by atom type, then I did a Kabsch-optimised rotation. It works pretty well. Hope this helps, I can even provide the piece of code doing it (but integrated in a bigger software)." First, thank You for the offer! And some thoughts to your comment: The idea to use an isotropic characteristic of the atoms like 'distance to a center' instead of introducing a standard orientation is great. Maybe three or four such values would be enough to get directly a unique isotropic characterization of each atom. I think of something like the distance to the "nuclear-carge^n" weighted geometric center. Like: center of order 0 (x_i = x-Coordinate of atom i, ...): Sum_[i->n] ( x_i + y_i + z_i ) / n center of order 1 (z_i = nuclear charge of atom i): Sum_[i->n] z_i ( x_i + y_i + z_i ) / n center of order 2: Sum_[i->n] (z_i)^2 ( x_i + y_i + z_i ) / n ... These centers would be easily accesible, uniquely defined, and (most often) different points, in the molecular coordinate system. And giving the distance of one atom to all these points would uniquly determine the atoms position. After that one gets two files with atoms in different ordering but with identical atoms having the same characteristcal distances. And now i would instead of trying out permutations (also called rotations) just canonically order the atom sequences (S0), this could be most easily done with a lexicographical sorting according to these characteristics. In this way one would yield directly two Permutation operators (P1,P2), one for each file (S1,S2) having the following property: P1*S1 = S0 = P2*S2 and hence S1 = P1^(-1) * P2 * S1 so P1^(-1) * P2 , is the Operator in question. In this way the task of trying out permutations could be shifted to the usage of standard search algorithms (sort). - just a thought - ~~Tom: I'm mainly concerned about inorganic molecules with about 20-40 atoms, so the smiles approach might not be the method of choice for me - if I understand correctly. Moreover I'm in a kind of permanent financial crisis ;-) Best regards Raphael From owner-chemistry@ccl.net Sat Jun 6 09:31:00 2009 From: "Mahmoud A. A. Ibrahim m.ibrahim1982%yahoo.com" To: CCL Subject: CCL:G: Different atom symboles Message-Id: <-39457-090606092818-29291-JzBuRS5Ru5KE5tbV9MHSfQ ~~ server.ccl.net> X-Original-From: "Mahmoud A. A. Ibrahim" Date: Sat, 6 Jun 2009 09:28:11 -0400 Sent to CCL by: "Mahmoud A. A. Ibrahim" [m.ibrahim1982!^!yahoo.com] Dear CCL users I was wondering, what is the difference between ?s ?d ?t and ?r ? these are atom symboles defined inside gaussian view (Guassian View --> Atom List Editor ---> symbole). I know just the first one, ?s is defined when you delete a terminal atom of a bond with keeping the bond. i.e. in case of C-H you deleted the H atom and keep the bond to give you C- ok, also what is the aim of these types? what functions they have? I took forward to hearing from you. Sincerely; M. Ibrahim From owner-chemistry@ccl.net Sat Jun 6 22:45:01 2009 From: "anessh cna aneeshcna . gmail.com" To: CCL Subject: CCL: unit conversion of force constant Message-Id: <-39458-090606050850-24972-QQFa5sCvBUKYVunZphKTMA . server.ccl.net> X-Original-From: "anessh cna" Date: Sat, 6 Jun 2009 05:08:46 -0400 Sent to CCL by: "anessh cna" [aneeshcna%%gmail.com] Dear CCL users, I am doing vibrational analysis for calculating the bond and angle force constant for developing force field parameters for an organic molecule. Can anyone help me in converting the force constant in N/m to kJ/mol/A**2 or kJ/mol/rad**2 ?. Here I am tring to calculate force constant from frequencies. Thanks in advance Aneesh aneeshcna . gmail.com