I did not know MZMine, so I looked it up. Its main application seems to be LC-MS data, not GC-MS. This most probably means that the software might have problems removing the column bleed (like MassHunter), which is the main baseline contribution in GC-MS.

If your baseline is high due to your samples, the baseline of your blanks should be lower. If that is not the case I would try to fix this issue instead of finding a software mitigation. What column are you using and what are your baseline masses?

For qualitative GC-MS analysis I use AMDIS, which is free and gives you "cleaned up" spectra for library searches. For quantitative analysis I look for masses with a high selectivity for my analytes, which may be different masses for the same analyte in different matrices (but usually it´s not). I would really like to learn, if MZMine or any other software can do this in an automated way.

We had a high noise issue in LC/MS, and after several experiments and investigations we figured out that it was due to water's conductivity . so try to see the blank sample's noise, if its also high, try to treat the samples with a better dilution solvents that can reduce the noise.

To my knowledge MZmine (2) is a software package for differential chromatography analysis with the main focus on LC-MS data. Peaks in GC-MS are more narrow, the baseline is often more noisy. Therefore it seems to be a good idea to do a smoothing before baseline correction and deconvolution of peaks. But you have also to choose the right "Peak Deconvolution" mode and also an appropriate "Minimum time span" for peak detection. This needs testing....

Hey David

Can I ask what your experimental setup is, i.e. Instrument, column, inlet temp, gradient program etc...

Hi Alexander,

I have seen that MZmine can be used for GC-MS as well as LC-MS, though i agree it might be trickier.

This 2014 study compared it with a few other automated tools for GC-MS and LC-MS use:

http://www.sciencedirect.com/science/article/pii/S0021967314010450

It looks like the baseline of my blanks is about 40% lower than the baseline of samples run at a similar time to the blanks. Is there a way to actually subtract the TIC of the blanks from the samples? or is that a bad idea?

Hi David,

I am not sure I would do any baseline smoothing and corrections. Mentioned deconvolution with AMDIS should work if you have a good resolution data (at least 9 scans per peak). I am sure you now that deconvolution algorithm works in the way that it search individual ions and recognise pattern where group of ions form a peak. That means that the background, many column bleed, is ignored in peak detection, and I think in area estimation too (but check). 

I hope this is helpful.

Hi David,

We are working with PARAFAC2 which does baseline correction, peak detection and alignment at the same time. PARAFAC2 analysis all samples in one workflow, which might be very beneficial in your case. Have a look at our homepage and let me know if you would like a demonstration.

http://msomics.com/data%20processing/index.html

Hi Morten, that looks interesting. Is it downloadable software? I can't see any download link.

Hi David, The PARAFAC2 algoritm is freely available in MatLab but will take some work to use. We have together with Bekzod Khakimov, Peter Bæk Skou and Rasmus Bro developed a graphical user interphase which enables the full processing of a sample set in less than two days of work. This graphical user interphase will be released as freeware in a stand alone version (no MatLab required) together with a publication. However, until then we either use the software on a fee-for-service basis or as part of scientific collaborations. We offer demonstrations of the software and tests you data for suitability free of charge.