New tools are filling the gap in our metabolomics capabilities.

Metabolomics can provide a better understanding of the state of cellular and biological processes at different stages of growth, under disease conditions, in response to stimuli, and even as a molecular signature for cell identification. The metabolome is thought to be the closest representation of phenotype and shows tremendous potential in applications such as understanding cellular processes in both normal and diseased tissues, environmental research, agriculture, biomarker discovery and personalized medicine. In spite of this, however, comprehensive data analysis and identification of the complex metabolite pool is still a challenge.

Sophisticated tools are needed to process metabolomics data. Credit: Kimberly Caesar/Springer Nature

One limitation is the sheer number of still uncharacterized metabolites — the small sizes of publically available spectral libraries, in comparison to the possible metabolite space, means that a high percentage of peaks remain unidentified in mass-spectrometry-based experiments. Another limitation is accurate statistical validation for large-scale spectral assignments. Again, due to the high metabolite diversity, generation of a comprehensive decoy set to estimate the false discovery rate, which has been successful in the field of proteomics, is not feasible. Available approaches often require parameters to be adjusted for each data set (Nat. Commun. 8, 1494, 2017).

However, progress is being made. Some of the new approaches for compound identification include integration of CSI:FingerID in SIRIUS4 (Nat. Methods 16, 299–302, 2019) for searching molecular structure databases, multi-level annotations using chemoinformatics strategies for carbon number determination and metabolite classification (Nat. Methods 16, 295–298, 2019), integrative metabolomics platforms such as MetaboAnalyst 4.0 (Nucleic Acids Res. 46, W486–W494, 2018), and metabolic reaction network–based annotation (Nat. Commun. 10, 1516, 2019). Creative applications and analyses include the use of metagenomics and genome-scale metabolic models to aid microbiome modeling (Nat. Microbiol. 3, 456–460, 2018) and integration of metabolomics data with metabolic pathways for identifying metabolites affecting particular phenotypes (Nat. Biotechnol. 36, 316–320, 2018).

The potential applications of metabolomics are only beginning to be fully explored. As researchers combine other -omics data and with more sophisticated study designs, the need for methods, both generalizable and specialized, will continue to grow. We look forward to further developments in the field.