Bioinformatic Analysis of Metabolomic Data: From Raw Spectra to Biological Insight

Abstract

Metabolites are at the end of the gene–transcript–protein–metabolism cascade. As such, metabolomics is the omics approach that offers the most direct correlation with phenotype. This allows, where genomics, transcriptomics and proteomics fail to explain a trait, metabolomics to possibly provide an answer. Complex phenotypes, which are determined by the influence of multiple small-effect alleles, are an example of these situations. Consequently, the interest in metabolomics has increased exponentially in recent years. As a newer discipline, metabolomic bioinformatic analysis pipelines are not as standardized as in the other omics approaches. In this review, we synthesized the different steps that need to be carried out to obtain biological insight from annotated metabolite abundance raw data. These steps were grouped into three different modules: preprocessing, statistical analysis, and metabolic pathway enrichment. We included within each one of them the different state-of-the-art procedures and tools that can be used depending on the characteristics of the study, providing details about each method’s characteristics and the issues the reader might encounter. Finally, we introduce genome-scale metabolic modeling as a tool for obtaining pseudo-metabolomic data in situations where their acquisition is difficult, enabling the analysis of the resulting data with the modules of the described workflow.