Occasionally users have told me that spectra they have collected on the Skaggs NMRs do not look as good when they take them back to their own computers. This could be because they did not process their data with optimised parameters. Or, since most users use Mnova, it may be due to real processing differences between Mnova and TopSpin. Here I examine a few spectra processed with both packages to see if there is a real difference between processing with Mnova and TopSpin.
The most obvious difference I have noted between data processed with Mnova and TopSpin is the noise in 1D spectra. The figure below shows an expansion of a 13C DEPTQ spectrum. The TopSpin processed data is the upper spectrum, and the Mnova processed the lower one.
There is little difference between the peaks, but the noise does appear different. To my eye, the Mnova noise does not look truly random. It looks as though it was generated to fill a specified range.
More concerning than differences in noise, though, are reports of peaks seen in TopSpin processed data but not in Mnova spectra. The figure below shows an expansion of the aromatic region of a HSQC spectrum. The TopSpin processed data is at the top and the Mnova below. Identical window functions and amounts of linear prediction in the indirect dimension were used in both packages. Contour intervals are the same in both plots. The threshold is roughly the same but could not be matched exactly as the packages scale the data differently. The TopSpin data appears to have slightly better resolution in the 1H dimension and slightly better signal-to-noise.
When I analysed this data using TopSpin I picked six peaks in the expansion shown above. In addition to the four obvious peaks at 6.23-141.9, 6.21-133.5, 6.07-139.5 and 5.81-129.8, I picked another at 6.06-129.4 and a sixth at 5.93-143.1. The sixth peak is very weak and I used other spectra to help convince me that there was a peak at this location. However, in the Mnova processed spectrum this peak is even less convincing.
While the Mnova data did not look as good as the TopSpin data the difference was not that great. I reprocessed the spectrum with Mnova's default parameters to see if this could cause the disappearance of peaks. The result is shown below. Here the resolution in the 13C dimension is not as good as the spectra above. The weak sixth peak is not visible, and even the fifth peak at 6.06-129.4 is dubious.
The answer seems to be that Mnova's default processing parameters do not always give the best results. This is probably true of any data processing package. It is worth taking the time to learn how window functions, zero filling, and linear prediction can alter the appearance of your data. You can probably get more information out of your data than you realise.
Acknowledgements
Thanks to Gisela Camacho-Hernandez and Jim La Clair for the use of their data.
It's worth mentioning that by default MNova won't apply the TopSpin-defined windows functions etc. There is a checkbox in the options to make this automatic.
ReplyDeleteThanks for the comment. I did not know this, but I may start recommending it. Thank you.
Delete