Fig. 1

Selecting a data integration strategy depends on the structure of accessible multi-omics data. (a, left) If multiple types of omics data (‘multi-omics’) are available for the same cohort of patients, vertical integrative analysis can be performed to combine information across data types. This integration can be achieved using a variety of multi-staged and meta-dimensional statistical approaches that identify disease subtypes, regulatory networks, and driver genes. (a, right) If the opposite is true and a specific type of omics data is available across a number of different patient cohorts, horizontal meta-analysis can be performed to increase statistical power and identify disease-associated perturbations. b In some cases, however, experimental data are only available for different omics data types from different cohorts of patients and neither vertical nor horizontal data integration can be performed. In these situations, integration relies on mapping data to common units (e.g., genes or pathways) and then either integrating transformed data or simply overlapping candidate sets. The software approach presented here (integRATE) utilizes desirability functions to transform and integrate heterogeneous data allowing for the prioritization of candidate genes for functional analysis