Re: [BlueObelisk-discuss] help needed -- SMILES and MMFF94 atom types
follow-up on this. Over the last few days I was able to extend this to fully validated MMFF94 minimization capability in Jmol with empirical rules. Here's a summary of what I just checked in. version=12.3.26_dev # code: adding empirical rules to MMFF94 calculation # # checkmm.spt;checkAllEnergies # checking calculated energies for 761 models # 1 COMKAQ E= -7.3250003 Eref= -7.6177diff= 0.2926998 # 2 DUVHUX10 E= 64.759995Eref= 64.082855 diff= 0.6771393 # 3 FORJIF E= 35.978 Eref= 35.833878 diff= 0.14412308 # 4 JADLIJ E= 25.104 Eref= 24.7038diff= 0.4001999 # 5 PHOSLA10 E= 111.232994 Eref= 112.07078 diff= 0.8377838 # 6 PHOSLB10 E= -93.479004 Eref= -92.64081 diff= 0.8381958 # 7 OHMW1 E= -20.78 Eref= -21.726902 diff= 0.9469013 # for 761 atoms, 7 have energies differences outside the # range -0.1 to 0.1 with a standard deviation of 0.06309618 COMKAQ -- BATCHMIN ignores 1 of 5-membered ring torsions for a 1-oxo-2-oxa-bicyclo[3.2.0]heptane -- MMFF94_bmin.log: WARNING - Conformational Energies May Not Be Accurate DUVHUX10 -- BATCHMIN ignores 5-membered ring issue for S-S-containing ring -- MMFF94_bmin.log: WARNING - Conformational Energies May Not Be Accurate FORJIF -- BATCHMIN misses four standard 5-membered C-C ring bonds -- MMFF94_bmin.log: WARNING - Conformational Energies May Not Be Accurate JADLIJ -- BATCHMIN ignores 5-membered ring for S (note, however, this is not the case in BODKOU) -- MMFF94_bmin.log: WARNING - Conformational Energies May Not Be Accurate PHOSLA10 -- BATCHMIN ignores all 5-membered ring torsions in ring with P -- (note, however, this is not the case in CUVGAB) -- MMFF94_bmin.log: WARNING - Conformational Energies May Not Be Accurate PHOSLB10 -- BATCHMIN ignores all 5-membered ring torsions in ring with P -- (note, however, this is not the case in CUVGAB) -- MMFF94_bmin.log: WARNING - Conformational Energies May Not Be Accurate OHMW1 -- H2O complexed with hydroxide OH(-) -- I don't understand (a) why the OH(-) bond has mltb=1, and even with that I am not getting the correct ro/kb for that bond from empirical rules. Still working on that I think SMARTS was definitely the way to go on this. -- Robert M. Hanson Professor of Chemistry St. Olaf College 1520 St. Olaf Ave. Northfield, MN 55057 http://www.stolaf.edu/people/hansonr phone: 507-786-3107 If nature does not answer first what we want, it is better to take what answer we get. -- Josiah Willard Gibbs, Lecture XXX, Monday, February 5, 1900 -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/___ Blueobelisk-discuss mailing list Blueobelisk-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss
Re: [BlueObelisk-discuss] help needed -- SMILES and MMFF94 atom types
follow-up on this. Over the last few days I was able to extend this to fully validated MMFF94 minimization capability in Jmol with empirical rules. Here's a summary of what I just checked in. Great! I think SMARTS was definitely the way to go on this. Well, I may co-opt your SMARTS back for Open Babel. ;-) OK, I'm stumped. How does one get to Rule d? Does anyone know if this code was ever tested? That clearly looks like a bug. I forwarded it to Tim, but I haven't heard back from him. Presumably the rule c should be an else if (…) instead of just an else. -Geoff --- Prof. Geoffrey Hutchison Department of Chemistry University of Pittsburgh tel: (412) 648-0492 email: geo...@pitt.edu web: http://hutchison.chem.pitt.edu/ -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ ___ Blueobelisk-discuss mailing list Blueobelisk-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss
Re: [BlueObelisk-discuss] help needed -- SMILES and MMFF94 atom types
Here's my final result. I'm satisfied; I'm pretty sure there is no way to reproduce the results of the OPTIMOL results exactly. In the cases below I'm convinced there are bugs in OPTIMOL -- mostly in not consistently recognizing 5-membered rings, but also there is something odd going on in the empirical bond parameter calculation. Bob # version=12.3.26_dev # # code: adding empirical rules to MMFF94 calculation # # checkmm.spt;checkAllEnergies # # checking calculated energies for 761 models # 1 COMKAQ E= -7.3250003 Eref= -7.6177diff= 0.2926998 # 2 DUVHUX10 E= 64.759995Eref= 64.082855 diff= 0.6771393 # 3 FORJIF E= 35.978 Eref= 35.833878 diff= 0.14412308 # 4 JADLIJ E= 25.104 Eref= 24.7038diff= 0.4001999 # 5 PHOSLA10 E= 111.232994 Eref= 112.07078 diff= 0.8377838 # 6 PHOSLB10 E= -93.479004 Eref= -92.64081 diff= 0.8381958 # # for 761 models, 6 have energy differences outside the range -0.1 to 0.1 # with a standard deviation of 0.05309403 # # a comment about empirical bond parameter calculation: # #// Well, guess what? As far as I can tell, in Eqn 18 on page 625, #// the reduction term and delta are zero. # #// -- at least in the program run that is at the validation site: #// OPTIMOL: Molecular and Macromolecular Optimization Package 17-Nov-98 16:01:23 #// SGI double-precision version ... Updated 5/6/98 #// #// This calculation is run only for the following three structures. In each case the #// reported validation values and values from Jmol 12.3.26_dev are shown. Clearly #// the r0 calculated and final energies are very good. subtracting off 0.008 from #// r0 would certainly not give the reported values. Something is odd there. #// #// bond red* r0(here/valid) kb(here/valid) Etotal(here/valid) #// --- #// OHWM1 H1-O1 0.03 0.978/0.978 7.510/7.51 -21.727/-21.72690 #// ERULE_03Si1-P10.0 2.223/2.224 1.614/1.609 -2.983/ -2.93518 #// ERULE_06N1-F1 0.0 1.381/1.379 5.372/5.438 1.582/ 1.58172 #// #// *reduction and delta terms not used in Jmol's calculation # # -- Robert M. Hanson Professor of Chemistry St. Olaf College 1520 St. Olaf Ave. Northfield, MN 55057 http://www.stolaf.edu/people/hansonr phone: 507-786-3107 If nature does not answer first what we want, it is better to take what answer we get. -- Josiah Willard Gibbs, Lecture XXX, Monday, February 5, 1900 -- Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/___ Blueobelisk-discuss mailing list Blueobelisk-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss