Hello. My name is Oscar Yanes.
I'm the scientific coordinator of Ciberdem
which is the Spanish National Network for the Study
of Diabetes, and I'm Assistant Professor at
the Universitat Rovira i Virgili in Tarragona, Spain.
The second part of the talk covers mainly the bioinformatic aspects of
the metabolomics workflow, involving
the pre-processing of the data and metabolite identification.
So once the samples have been analyzed by mass spectrometry or nuclear magnetic resonance (NMR),
we will have tonnes of a spectra that we need to process for data analysis.
Converting raw spectrometric data into
biological knowledge, has become the major bottleneck for metabolomics.
As you can see here in this slide going from raw spectral data -
either mass spectrometry data or NMR data, to biological information, to the name
of the metabolites, and the association of these names and their structures to
particular pathological diseases, is the main problem in the field right now.
The reason is twofold.
In one aspect, it is because of
the large physico-chemical diversity of compounds that I introduced at the beginning.
But if we go to the next slide,
I'm sure that anyone listening to my presentation, has been in
a discotheque or pub, with many people talking at the same time,
and if you try to distinguish the different dialogues in this room;
it becomes very difficult to separate the different dialogues, coming from each person.
This is why it's called the cocktail party problem.
In metabolomics we have a similar problem.
Instead of dialogues, we have peaks, and then we have
a huge overlap of peaks in our spectra.
In order to identify the structure and the name of these metabolites,
we need to separate or extract the peaks, belonging to each metabolite from the spectra.
Here, you can see what I'm referring to as the cocktail party problem in metabolomics.