An approach integrating several high-throughput datasets such as transcriptomes, proteomes and metabolomes.Multi-omics is the practice of measuring several different classes of molecules from the same biological sample and analysing them together. Each 'omics' layer captures one level of biology: genomics looks at DNA sequence, transcriptomics at RNA transcripts, proteomics at proteins, and metabolomics at small molecules. A single layer gives a partial view; combining them lets researchers see how a change at one level propagates to another.
It matters because biological function is not determined by any one layer alone. A gene may be present but not expressed; an RNA may be made but not translated into protein; a protein may be present but inactive. By measuring several layers in the same samples, multi-omics can distinguish these cases and point to the actual drivers of a condition. It is widely used in cancer research, where tumours differ in their DNA, RNA and protein profiles, and in studies of how organisms respond to treatments or environments.
The approach explains how to connect genotype to phenotype: it links a genetic variant to the transcripts, proteins and metabolites it affects, and it can reveal regulatory relationships that no single dataset shows. The main challenges are practical — different layers require different sample preparation, and integrating datasets with different scales and noise levels demands careful statistics.
The term multi-omics emerged in the 2000s as high-throughput sequencing and mass spectrometry made it feasible to measure several molecular layers at once. Early integrated studies combined transcriptomics and proteomics, and the approach expanded as metabolomics and other methods matured.
Samples are split so that each omics layer is measured from the same biological material, then the resulting datasets are integrated computationally. Integration can be done at the level of features, by linking genes to their transcripts and proteins, or at the level of samples, by finding shared patterns across layers. Statistical methods look for correlations and causal relationships between molecules across layers, and machine-learning models are often used to combine them into a single predictive signature. The goal is to identify a small set of molecules that explain the observed biology rather than listing thousands of changes in isolation.
It is used in precision medicine to match patients to treatments, in cancer research to classify tumours by their molecular profiles, and in microbiology and plant science to study how organisms respond to their environment.
The word 'omics' is a back-formation from 'genomics', and the proliferation of such terms has led to the tongue-in-cheek label 'ome-omics' for the study of omics itself.