Spectral prediction

Prediction of a spectrum corresponding to a postulated structure is a key element in structural confirmation or elucidation. 

In large molecules, manual addition of all the SCS can be tedious and a computer programme was soon developed in order to do this automatically and others followed.  

A much more common and probably more successful approach to the prediction of shifts are those derived from databases. Here, a database manager searches a database of structures for those containing sub-structures within the postulated structure. The key elements necessary for this approach are  

collections of H shifts exist which can be searched manually e.g., methine and methylenes. More usefully, proton chemical shifts have also been the subject of SCS or additivity rules. Prestsch and co-workers devised 

Advanced Chemistry Development Inc. has built a sizeable proton chemical shift database derived from published spectra (most commonly in CDCI3 solution). Their H NMR predictor programme accesses this database and allows the prediction of chemical shifts. Whilst this software takes account of geometry in calculating scalar couplings, in predicting chemical shifts it essentially treats the structure as planar. It would therefore seem doomed to failure. However, if closely related compounds, run at infinite dilution and in the same solvent, are present in the database, the conformation is implied and the results can be quite accurate. Of course, the results will not be reliable if sub-structures are not well represented within the database and the wide dispersion of errors (dependent on whether a compound is represented or not) can cause serious problems in structure confirmation (later). ACD are currently revising their strict adherence to HOSE codes for sub-structure identification and this will hopefully remove infrequent odd sub-structure selections made currently. 

---

Ball JW, Jurs PC (1993) Automated selection of regression models using neural networks for 13C NMR spectral prediction. Anal Chem 65 505... 

Recently, Okuyama et al. succeeded to prepare single crystals of some azobenzene amphiphiles and decided molecular and aggregation structure of single crystals [14-19], The spectral prediction of the chromophore orientation in the bilayer assemblies were very consistent with the X-ray structural analyses of the single crystals. 

Tandem mass spectrometry has been used to demonstrate that M+ as well as MH+ of substituted A-(ort/zo-cyclopropylphenyl)benzamides isomerizes before the fragmentation, with formation of 3-aryl-1-ethyl-lH-benzoxazines and 5-ethyl-2-oxodi-benzoazepines (Scheme 5.14). The methyl group in /V-[ortho-( 1 -methylcvclopropyl )-phenyl]benzamides quenches the latter process, leaving the formation of benzoxazines as the only cyclization reaction. A subsequent chemical experiment in solution confirmed the mass spectral predictions [24]. A similar study confirmed the analogy of cyclization of substituted A-(ort/zo-cyclopropylphenyl)-A -aryl ureas and N- ortho-cyclopropylphenyl)-A -aiyl thioureas in the ion source of mass a spectrometer and in solution [25]. 

Automatic spectral processing, spectral analysis, spectral prediction and structural confirmation elucidation have been reviewed. and two-... 

Simpson, A. J., Lefebvre, B., Moser, A., Williams, A., Larin, N., Kvasha, M., Kingery, W. L., and Kelleher, B. (2004a). Identifying residues in natural organic matter through spectral prediction and pattern matching of 2D NMR datasets. Magn. Reson. Chem. 42,14-22. 

The (TTF" )2 dimer that we have used to test our spectral predictions appears to be well suited as a prototype system. In fact,the electron transfer interaction between TTF " cation radicals is strong,and the high symmetry of the dimer (effectively D2fc) prevents any mixing of the CT and the LE excitations. We emphasize that the selective RRS enhancement of the intramolecular phonon modes is a distinctive feature of the present model and test case. It occurs under the restrictive conditions that (i) the dimer is symmetric (ii) the CT states do not mix with localized molecular excitations. When any of these two conditions is broken,the intramolecular modes can participate in the RRS enhancement. 

Before giving a few examples of applications of dimeric models, and for the sake of comparison, it is worth to derive a few spectral predictions for the case of regular stacks. 

Combined with some simple symmetry arguments applicable to infinite-stack systems with regular structures, the dimeric models provide us with a set of selection rules and spectral predictions that make infrared and Raman spectoscopy a powerful tool for structural investigations. These methods become rather unique if one considers the widespread availability of the required instrumentation and their applicability to disordered and amorphous materials like, e.g., thin films and Langmuir-Blodgett films which are more apt for applications than crystalline materials. 

Spectral prediction is required especially in the case of characterization and structure elucidation of large complex molecnles, such as natural products [30], A complete one-to-one correspondence for the assignment of the peaks in the spectra is not possible from the experimental spectra. Prediction is also required in the case of mechanistic nnderstanding for synthetic organic chemistiy. Many methods have been developed to predict spectrum, given stmctural information. 

Models Using Neural Networks for i c NMR Spectral Predictions. 

The vibrational spectrum of benzene has been discussed many times (29, 102-108) and, except for the frequencies of representation J52 , has been assigned unequivocally (109,110). Interestingly the symmetry D was not immediately established from the vibrational spectrum, since the latter exhibits some pecularities, such as Fermi resonance of combination bands, etc. This was not at first recognized. From the spectral predictions of Table XXVIII we now know that four infrared- and seven Raman-active n.v. should be observed for the planar De molecule. 

---