Deconvoluting Noisy Data

If you are deconvoluting a noisy data set there are some things that you can do to ensure that you are extracting the most accurate information from your data.

For data with high baseline noise increase the baseline factor:

If your data has a lot of baseline noise where the real signal is riding on a "hump", increase your baseline factor to remove this noise. The spectrum below is an example of a high baseline noise spectrum before baseline removal was applied.

After applying baseline removal with a baseline factor of 1, the following spectrum was obtained.

Use comprehensive deconvolution mode:

Noisy data often means that the spectrum contains a complex mixture. Comprehensive deconvolution mode is better for complex data sets. With comprehensive mode turned off, the algorithm will remove components from the data as they are found, which tends to favor the most abundant components in the spectrum.

Adjust the Noise Threshold:

ZNova may not be able to accurately determine the noise level for high noise spectra. Therefore, in these cases it is often desirable to manually set the noise threshold settings, particularly if you are interested in low level components. You can try changing the S/N setting from the default value of 2 from the auto threshold setting.  Lower the S/N setting if you feel that valid peaks are being discarded from deconvolution.  Increase the S/N setting to eliminate artifact peaks from deconvolution of noise peaks. Alternatively, you can set the threshold to a % relative intensity level.   The ESI spectrum from above was deconvoluted with a relative intensity noise threshold of 1% to obtain the spectrum below:

Mass Range:

Noisy data sets can slow down the ZNova algorithm considerably, simply because the difference between signal and noise may be not be easily discernable and ZNova has to process many more data points. You can help speed up the process by limiting your input and output mass ranges to the desired regions.