Metabolomics in human urine

[dRima]

Hi, all!
I'm a beginner in metabolomic studies, but now I have the task - metabolic profiling of human urine, to search for target and very improtant - non-target metabolites. In general, to find the changes in profile.
The samples were derivatized using BSTFA and made GC-TOF experiment (High Resolution TOF).

I got quite a rich of profile of the total ion current with the major and minor components. As I understand it, the next step is deconvolution profile. Here my question to you:

How should I correctly interpret the data obtained after deconvolution? Because, after deconvolution i have a large number of unknown components, many of which have 3 or 5 mass in the spectrum and certainly i can not consider such results as a "possible compound".
What criteria do you choose? What minimum mass in mass spectrum must be after deconvolution, for believe further confirmation of the spectrum in the NIST library (for example). If you can, please share your experience!

P.S. Or, i can work without identifying the compounds in the profile and should be performed just comparison of the two profiles and then find and interpretation only peaks, which was different?

Thank you!

[Peter Apps]

Hi, all!

P.S. Or, i can work without identifying the compounds in the profile and should be performed just comparison of the two profiles and then find and interpretation only peaks, which was different?

Thank you!

Certainly you should focus only on the peaks that differ between the samples - these are the biomarkers that you are looking for. If you look in enough detail you will find dozens or hundreds of differences - begin with the biggest and most obvious ones. Trying to identify everything will take for ever.

Good luck.

Peter

[dRima]

Thank you, Peter!
This is very important reply. Of course major peak, has a high ratio between S/N. And these peaks, I can easily distinguish and thereafter to identify, by means an accurate mass.
But what is correct with minor peaks? You make an emphasis on the peak-to-noise ratio? (for example, if the peak of the intensity ratio above 3/1, this can be regarded as a peak that requires study). If less than 3/1, then we can say that it is mistake, which allocated after peak deconvolution procedure.

And is there any principles? This will help in the future discard a large number of peaks that do not need to study, how i understand.

With best regards!
Dmitry

[Peter Apps]

Hi Dmitry

Look only at the differences between samples. Initially look only at compounds that are present in one class of samples and absent in another, then those that are , say 100 times different in concentration, and so on.

Among the differences, start with those that have the biggest peaks and richest spectra, since they are the easiest to identify.

When you a dozen or so differences with identified peaks, then look at whether the differences make biological sense. The smaller the peak and the smaller the differences, the less interesting it is likely to be biologically. It is a waste of time trying to identify hundreds and hundreds of tiny peaks because, when TIC peaks or single ion peaks are close to the noise level deconvolution is not repeatable between samples, and apparent differences between samples are most likely to be random noise in ion signals.

If you have no clear differences between samples then you are possibly looking at the wrong class of molecule, or the experimental manipulation of the organism actually makes no difference.

Peter

[dRima]

Peter!

Thanks for your complete answer! It is very helpful to me!
I think I understood the idea of interpreting the data!
If I have any questions, I will also write them here in this topic.

Thank you again!

[GOM]

Hi Dmitry

I would agree with the above comments.

Can you disclose your reason for profiling? (e.g. Cancer detection?)

I have in the past carried out GCMS/olefactory(sniff ) analysis on a sample of my manager's morning urine - after an hour of sniffing the eluting peaks I felt quite ill

Interestingly, in this case I was looking at human excretions for a particular malodour - it turned out out that the human nose is so sensitive to this thiol (for good reason) that the sample had to be concentrated 10000 times to be able to detect it instrumentally - so sometimes biologically important molecules can be present at very low levels

Don't let that put you off - go for the above recommended comments initially

Cheers

Ralph

[dRima]

Hi, GOM!
This is absolutely not a secret! But the work is quite interesting and new for me. All the time we worked the determination of trace amounts of pesticides in agricultural products. So, always my group had the work with target analysis.

Now our university has a system of HRT, and for us come the task to study the system for further profiling of urine. For find difference in organic acids and other different compound. For my team collected two urine samples of adult and child, and asked try to find differences. It's a simple test, for further movements.

Therefore, I have such a general and simple questions. I understand starting point for data collection and process preparation of TIC for the deconvolution, but not much it was not clear how to work with the results after deconvolution.

With best regards!
Dmitry

[Peter Apps]

Hi Dmitry

The differences between adult and child urine are so large that you should see them clearly on the total ion chromatogram, you do not need to deconvolute the data.

You only need to deconvolute if the peaks that distinguish between the samples are overlapped by other peaks.

It sounds as if you are re-inventing the wheel. There is a huge amount of literature on urine profiling, and a massive online metabolomic database http://www.urinemetabolome.ca/

Peter