At 12:14 16/09/2007, Christoph Steinbeck wrote: >peter murray-rust wrote: > > I've had a long talk with Henry. Essentially his position is that he has > > a method for calculation which gives significantly increased accuracy in > > an automatic fashion. What I would suggest that Nick does is to use > > Henry's method on a carefully selected subset of NMRShiftDB. > >In fact, I should have stated more clearly that I'm very much in favor >of having Nick do NMRShiftDB G03 calcs using Henry's method. The stuff >that we've done is just supplemental.
Ack > > Stefan says it isn't available. In any case is it Open? I think Henry > > would contend his method would be an advance. > >The data are open and partly loaded into NMRShiftDB. I've written to >Warren again (did this a couple of times already) to get the rest. >But again, this is supplemental - let's go for your plan. Ack > >> I myself have been calculating about 2000 Gaussian spectra for > >> NMRShiftDB on a 16-node cluster which came for exactly this purpose with > >> my NMRShiftDB grant. > > > > Is this published? We don't want to pre-empt you. > >No, it is not published and I don't think it will. There was no rigorous >selection of a reasonable starting geometry in my protocol, so I would >have difficulties to justify my data. > >Ack >No - our 3D generator is broken at the moment. WONDERFUL - since these will be rigid and therefore almost all ring-based crystal eye will be ideal. > >> > The immediate difficulties we can see are: > >> > * not knowing the stereochemistry. *** WHAT IS THE POSITION IN > >> > NMRSHIFTDB? We can filter out anything that has more than one > >> > potential stereocentre. > >A lot of structures with stereocenters in NMRShiftDB have undefined >stereochemistry. It would be good to generate all of them in 3D and do >the calculation (in case of E/Z and diastereomers). Excellent. A really good use of Nick's infrastructure. The problem with computing geometries of crystal structures is that both the crystallography and the computation are very good. So the main result tends to be that they agree. That's valuable but undramatic. In this case the data (e.g. connection tables) are much less reliable. > >> > * assumptions about completeness of data in NMRShiftDB > >Not sure what that means. I think I mean "does every structure have all the peaks? is every peak assigned?" > >> > * generating initial 3d coordinates. There are several simple > >> approaches: > >> > - use CDK - I think Christoph had something here? > >> > - use the 2D coordinates for "flat" molecules > >> > *** NEW APPROACHES - MUST BE BLUEOBELISK > >How about this newly discovered "RDKit" (or so)? See Noel's blog. >Seems to have a BSD license 3D generator. Good. We'll also use crystaleye >Regarding the workup of the data: I would like to train our shift >prediction machine learning methods with the GIAO data and see how that >compares to the experimental data. Excellent. Suggest you and Nick liaise... Peter Murray-Rust Unilever Centre for Molecular Sciences Informatics University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK +44-1223-763069 ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Blueobelisk-discuss mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss
