[Boteachman]

Hi All,

TL;DR: I'm looking for a way (or simply inspiration on how) to predict molecular structures (isomers) given a chemical formula and it's corresponding raw mass spectrometry data.

I'm currently working on my final year thesis which is around building a model to identify and predict lipids from mass spectromtry data in an untargeted approach. Currently, my model outputs potential chemical formulas from a given isotope pattern extracted from raw MS data. Given these formulas, I would like to then predict the possible isomers.

Since I can find many tools that can predict retention time given a molecular structure, my initial idea was to predict retention times for all isomers of a given chemical formula and match it up with the observed retention time. However, the problem is generating these molecular structures. I tried generating all possible isomers given a chemical formula (using tools like MOLGEN and MAYGEN that I found online), however this was computationally very expensive and isn't feasible.

What will probably end up being done is that I will match the observed retention times with predicted retention times using lipid/metabolite databases from online. However, I would first like to see if there's a way to predict molecular structures, or even just possible structural features, given a chemical formula, observed retention time and observed isotope pattern from raw MS data. From my preliminary search there's no exicting tools that do this so I was hoping that someone would have some useful information to share on this. Please let me know if you do!

Thanks.