Chromatography Forum

First off, I am not a chemist but a biologist with a metabolomics project and I could really use your help!

I have done an untargeted metabolomics project using LC-MS. I have finished the data analysis, and I am finding that many peaks look good, but others leave something to be desired.

One feature in particular matches to a metabolite with great biological relevance in my system. Obviously additional experiments need to be done to confirm the identity, but I am wondering how to address the fact that the peak shape is horrible in my thesis. This feature elutes at the beginning of the run and produces a broad, flat peak.

I've uploaded the peak in question as well as an example of a good peak from the same dataset.

I would say that the first peak elutes too early in the run, so any quantification is absolutely unreliable. This is easily explained: if the mass is correct, the peak represents P-(Hydroxymethyl)-phosphonic acid, which will be unretained in reverse phase. Thus, the difference in area could be due to many factors other than a real difference in concentration: co-elution of salts, signal suppression at the start of the run, unreliable peak shape, etc...

To confirm its quantification, one would need to run the relevant samples with a HILIC or ion-exchange method.

I advise you not to draw any conclusion on this analyte with the current method

Thank you for your reply. I was wondering if it is possible that this compound may be a potassium adduct? I used MassTRIX to probe several databases (KEGG, LipidMaps, HMDB, MetaCyc), and I got a hit for several M+K39 ions. The compound I was interested in was actually hydroxyacetone.

I have seen some papers that used the M+Na and M+K ions for identification even without an M+H present, but I am wondering if that would be considered bad practice.

Some molecules give a stronger metal adduct peak than the protonated peak, so it's fine to identify them with the potassium adduct.

But this won't change the fact that whatever you are seeing here is not retained on the column, and therefore should not be considered for any statistical analysis.

If you're interested in things like hydroxyacetone, LC-MS is probably not the best approach. Not only will they not be retained on typical reverse phase columns, but they will also, to a large extent, be evaporated and pumped away like the solvent. Hydroxyacetone has a boiling point around 150 degrees, and your spray chamber is probably well above this. Try GC-MS, if you can.

There are probably quite a few things with that mass; for example propionic acid as a K+ adduct.

I agree with Carlo entirely, that injection peaks are unreliable.

Note that peak-finding software is often quite bad at recognising real peaks, which is why you were right to check visually, and why your instincts were correct. If the peak isn't "peak-shaped", don't trust it.

There are some decent websites, databases and documents of background ions, which can be useful, especially for low-molecular-weight things in injection peaks, as well as sustained background ions that may appear occasionally as "peaks" when the software has been over-enthusiastic.