- Version
- Download 0
- File Size 1.29 MB
- File Count 1
- Create Date May 2, 2024
- Last Updated May 8, 2024
Metabolite annotation is one of the most important bottlenecks in metabolomics data analysis. While genomics and proteomics are considered as massive analysis techniques, metabolomics cannot be considered as such. This is because metabolomics can measure a large number of metabolites in biological samples, but the identification of these metabolites is an imposing bottleneck where, routinely, only a few metabolites are typically identified, i.e., annotated. One of the main reasons that hampers the analysis of metabolomics data is that mass spectrometry does not generate a single signal per observed metabolite, but it generates multiple redundant signals per metabolite. Associating all the signals that stem from each metabolite is needed for its annotation and subsequent identification. This also allows transforming a high-dimensional list of signals (multiple variables per metabolite) to a list composed of only one quantitative variable per metabolite like in proteomics or genomics data. Computational strategies have been proposed to overcome this limitation but the complexity of mass spectrometry makes the automation of this process prone to errors. Alternatively, some algorithms exploit mass spectrometry data in reference libraries to overcome this problem, but these algorithms are limited to the annotation of metabolites contained in these libraries, which hampers the wide adoption of these strategies. Here we report a new algorithm to expedite this process without the need of reference libraries, thus reducing the mass spectrometry data complexity and facilitating the annotation of metabolites. In addition, we leverage the information from multi-omics data to provide more robust metabolite annotations based on machine learning multi-omics integration. Our advanced workflow will improve the annotation rate in metabolomics studies allowing gaining more insights into the metabolism of plants, animals and environment.
Download