Ranking of Structure Proposals by their Predicted CNMR-spectra
Structure elucidation can be done by application of isomer generation. In many cases, e.g. with proton-poor compounds giving a small number of correlation peaks in 2D-NMR,
the isomer generator program tends to create a very large list of structure proposals. Ranking of the structure proposals contained in such a large list of possible
structure candidates can be done by prediction of their CNMR-spectra and comparing these predicted values against against the experimental chemical shift values. The difference can
be used to sort the structure proposals.
Necessary input data:
Your email address - email must be validated before first usage
A SD-file holding up to 10,000 structure proposals - file will be recalled via http, molecules must end with "M END" and "$$$$", any SD-tag allowed
A file holding exactly "n" chemical shiftvalues for the isomers, each having "n" carbons - file will be recalled via http
Multiplicity can be given optionally S = C, D = CH, T = CH2, Q = CH3, O = S or T, E = D or Q )
You prepare all files on your local computer, you make these files available for http-transfer
The CSEARCH-Ranking-Engine picks up the files you have defined in the input-mask
Every combination of SD-File, Peaklist and email can only be calculated once
That's it !
You can try the following example:
Your email-address
nmrpredict.orc.univie.ac.at/rank/demorank.sdf
nmrpredict.orc.univie.ac.at/rank/demo.pkl
The resulting data contain:
A webpage holding the sorted hitlist
The best 3 structure proposals will be automatically sent to the CSEARCH-Robot-Referee for a detailled analysis
The given experimental peaklist is used for a "Spectral-Similarity-Search"
Example for a correct SD-file: Click here; File is NOT from an isomer generator program, just an example showing how much a SD-file can be distorted !
Example for a correct peaklist: Click here