Troubleshooting a negative result

What happens if I get a negative verification result?

There are several reasons to get negative results in the verification. Let´s take a look at some of them

1.All or most of the NMR tests fail with bad scores: this could be an indication of either the wrong structure, a very poor quality spectrum or solvent misrecognition, which will affect most tests very significantly. The recommended Operating Procedure is to check the solvent recognition first. To do so, make sure you are in ´Active Spectrum´ mode and that your graphical properties (Home/Properties/NMRSpectrum/Peaks/Curve) has the ´Peaks´ box ticked. This will display colour coding for the autoediting carried out by the software on the spectra. You may identify this way that a solute peak has been considered as solvent or viceversa.

In a case in which this happens, this can be corrected by right clicking on the peak label (pointed at by the arrow on the figure) and selecting Edit Peak Type, then marking this peak correctly (in the case above would be marked as compound and with the Flags unticked…)

Once GSD/AutoEdit has been modified, you can just re-run Verification and your GSD modifications will be respected. In this case, after running again, the result is totally different, as can be appreciated in the figure below:

2.Should you find that the score is reasonable but that some atoms are showing assigned in red and that either the PBM or NN tests have low values, then this could be due to an inaccurate prediction or to an atom which is not where it is supposed to be (in the latter case this would be a true negative result). In this case, check the ´Worst Multiplet´ and ´Worst Location´ feedback, as this might indicate where you should look first. In the case of the figure below, ´Worst Location´ is 9.49 ppm, and there we can see that there is a peak, which is being assigned to atom number 11, but the predicted value for atom number 11 is at 8.75 ppm. The user can decide here whether it is reasonable that the prediction is poor and that the real value of atom 11 could indeed be 9.49.

In this case, the user can grab the prediction label (the same could be done with the rectangles in a HSQC) by clicking on it (the number 11 label on the blue spectrum) and moving it to the 9.49 ppm position. Once this has been done, the verification can be carried out again, making sure that the ´Redo Predictions´ tickbox in the verification dialog is unticked. This will allow the user to see what the result of the verification would be in case the prediction was being made at 9.49 ppm, rather than 8.75 ppm. In this case, as the figure below shows, the verification ´Score´ has improved from 0.56 to 0.82 and the atom which was badly assigned (colour coded in red on the molecule previously) is now well assigned. Should the user decide that the presence of atom 11 at 9.49 ppm is possible given its chemical environment, then this is a likely indication that the reason for the previous lower score was simply a poor prediction.

3.A low result in 1H Nuclides Count could also be indicative of poor integration due to either impurities or a very short relaxation time. It is to be noted that a very short relaxation time can affect integration and that poor integrals will have a negative impact on the verification result. This can be thought of as a false positive but in fact the software is reporting a problem or incompatibility in the data with the proposed molecule.