Blob analysis - chemometrics

[nlim]

Dear All,

In the current version of GC image, there is a possibility to analyze blobs through PCA. Are the raw data of the blobs based on m/z signal in the whole blob area or the peak area only? Is there a reference documentation on how GC image does the preprocessing and PCA on the raw data? Also, is there a way to export the m/z-intensity raw data array (RT1, RT2, m/z intensity matrix) of the chromatogram to process it with other chemometric tools (e.g. cluster analysis, PLS, nonlinear regression methods)? I have noticed that exporting it to a simple CSV will only have the TIC information and not the m/z intensities. On the other hand, the mzXML file does not export metadata based on 1st and 2nd dimension retention times. Any tips and suggestions are very much appreciated.

Thank you!

[geschd23]

Are the raw data of the blobs based on m/z signal in the whole blob area or the peak area only?

The spectra used to compute PCA are based on your Search Library settings (Configure > Settings > Search Library). With the Peak MS type selected, only the spectra from the blob peak locations are used. With Blob/Area MS selected, all spectra within each blob are summed.

Is there a reference documentation on how GC image does the preprocessing and PCA on the raw data?

The documentation for Blob Analysis Tool can be found at https://www.gcimage.com/gcxgc/usersguid ... alyzeblobs.

Also, is there a way to export the m/z-intensity raw data array (RT1, RT2, m/z intensity matrix) of the chromatogram to process it with other chemometric tools (e.g. cluster analysis, PLS, nonlinear regression methods)?

We don't currently have an export format that matches your description exactly. Another possibility would be the Text MS format (File > Export Image > Export Image as Text MS). It includes the spectrum string of every data point formatted as "[m/z],[intensity];[m/z],[intensity];...", though it does not include the associated retention time.

[qtao]

For exporting the m/z-intensity raw data array, it may also be done with an Export Image plugin.

Contact support for more details.

[qtao]

For mzXML, the software does write out the 1D retention times. You may convert them back to 2D retention times if start time and modulation period are known. For example,

Code: Select all

RT2 = (rt - start time) % modulation period
RT1 = rt - RT2

Note that the above does not work for variable modulation.

[nlim]

Thank you very much for your responses.

I have a follow-up question (though this might be for another forum topic).

I am exploring the m/z intensities (baseline corrected in default MS mode) for use in a prediction model. However, I noticed that some of m/z intensity values are in negative values. Is there a way to restrict negative values and keep lowest intensity value to zero? Is there a work-around for this? Ideally, I plan to explore m/z intensities of blobs as input variables in the model and zero value would mean as not detected.

[qtao]

Raw and unprocessed intensity values for mass spec should be positive. Negative intensities usually result from some baseline correction or subtraction. By default, our software performs baseline correction on all MS channels. So you will see negative intensities.

You may filter negative intensities in your own program, or choose performing baseline correction on only TIC (which would leave raw intensities alone). If you choose to filter intensities, note that positive values do not always mean something detected. Some mass spec records background signals way above 0.