25 Meta-dimensional omics integration

In meta-dimensional analysis all omic datasets are analysed in a single, simultaneous analysis. This kind of approach typically avoids using domain knowledge-based procedures to independently reduce features in single omic datasets, and aims at integrating multi-omic datasets in their whole complexity. Meta-dimensional integration methods can be grouped following several criteria but here we briefly summarise the classification first coined by Ritchie et al. (2015) [57] and recently reviewed by Reel et al. (2021) [58] (we refer interested readers to those publications for a more in depth treatment of the topic), which classifies the methods into concatenation-based, model-based and transformation-based integration methods. The three kinds of integration methods can be used for unsupervised and supervised analysis of multi-omic data, including classification and regression tasks.

Contents of this section were created by Iñaki Odriozola and Antton Alberdi.


57. Ritchie MD, Holzinger ER, Li R, Pendergrass SA, Kim D. Methods of integrating data to uncover genotype–phenotype interactions. Nat Rev Genet. 2015;16:85–97.
58. Reel PS, Reel S, Pearson E, Trucco E, Jefferson E. Using machine learning approaches for multi-omics data analysis: A review. Biotechnol Adv. 2021;49:107739.