WWMM calculation of spectra (proposed)

Jean-Claude Bradley has offered us the spectra that come out of his Open Notebook projects…

1. Jean-Claude Bradley Says:
   October 11th, 2007 at 1:32 pm ePeter – it is great to see you contributing to open NMR data processing.
   For a source of raw spectra, we have been and will continue to upload NMRs of our starting materials and isolated products to ChemSpider.
2. http://www.chemspider.com/spectra.aspx
   and for example benzaldehyde:
   http://www.chemspider.com/Chemical-Structure.235.html
3. I think a lot of the errors in the literature are due to poor quality spectra and improper referencing. It is really useful to verify that the TMS peak is indeed exactly at zero and that phasing was done properly. JSpecView used by us and ChemSpider allows people to expand any part of the spectrum and integrate at will, which is very handy for measuring fine splitting patterns, including J constants in H NMR spectra. The problems with JSpecView and the new Java version have been resolved by Robert Lancashire and fixed on ChemSpider.

PMR: it would be very nice to make this completely automatic. I’ve been talking with Jim. You will need to have a CML files with 3D coordinates and the spectral peaks linked to the carbon atoms (we’re only doing 13C at this time). Have a look at something from NMRShiftDB. Then it is automatic. If you don’t use CML we have a converter for the spectra (from JCAMP) and you can use JUMBO, CDK or OpenBabel for Mol2CML or whatever. We don’t do 2d-3d conversion and this is a problem for the Blue Obelisk at present.
You have to include the solvent. Christoph may be able to add more ideas.
I don’t know how NMRShiftDB did the assignments – I assume someone put them in by hand at time of submission. I am not sure whether JSpecView allows this assignment yet.
The current plan is for something like:
– human uploads 1 (or more) CMLSpect files (3D coords + assigned spectrum) at random time.
– job is given unique ID. molecule is checked for validity and uniqueness. Molecules that are too floppy or too large are filtered out.
– new “directory” is created with uniqueId and submitter is informed of URI. Format to be determined but something like =http://wwmm.ch.cam.ac.uk/nmr/jcb/1/ . Input file is /input.cml
– job is converted to gaussian input /input.gau and added to list of josbs-to-be-submitted
– every morning the submitted jobs are cron’ed into Condor.
– when a job finishes it writes /finshed.atom and /output.cml. (It may also create output.html, output.svg to display the results for that molecule on a y = x graph). The finished.atom (atom=Atom Publishing Protocol) alerts the submitter.
Comments are welcome. Jim and I have to fix it.