Integrating genomics, transcriptomics, proteomics, metabolomics, and other omics

Matched profiles from high-throughput genomics, transcriptomics, proteomics, and metabolomics provide complementary views from different angles. Integrative multiomic analyses can therefore help us gain an improved understanding on the molecular architecture of complex diseases such as atherosclerotic cardiovascular disease (ASCVD). There is much hope that the increased collection of multiomics profiles will lead to improved patient stratification, biological insights, robust biomarker discovery, and drug target identification. Fully integrated analyses have remained challenging, however, limiting routine use both in research and clinical applications. We here discuss both general challenges as well as considerations specific to ASCVD, where for instance access to relevant biological tissues may not be possible. Sex and ethnicity are often not sufficiently taken into account and many datasets are from samples of European ancestry. Both limit sample representativeness. Preanalytical factors and genetic and environmental influences, including diurnal variation, effects of aging and use of medications such as statins can feature as variables of a study or act as confounders. In general, different omic layers show different characteristics in their biologically relevant signals, their measurements bias and noise. Data are high-dimensional and sparse, increasing the risk of overfitting. We discuss current data integration approaches, including statistical, machine-learning, and network-based strategies, as well as potential applications. Moving forward, integrating molecular data with clinical, lifestyle, and imaging information, improving visualization, and adopting data deposition and federated, privacy-preserving data sharing will be pivotal. Comprehensive benchmarks, systematic inclusion of sex and ancestry, and further development of data integration approaches will help move multiomic ASCVD research from promise to clinical impact.