Multiset sparse partial least squares path modeling for high dimensional omics data analysis

摘要

Technological developments have enabled the measurement and storage of a plethora of biomolecular data extracted from various omics domains, such as data from the genome, epigenome, proteome or metabolome. It has become common to measure hundreds of thousands of biomolecular variables. To explore biological pathways across multiple omics domains, which might be associated with phenotypic (e.g. disease) outcomes, a natural research direction is to simultaneously analyse these omics domains. Complex diseases, such as obesity, diabetes, and schizophrenia have genetic architectures that involve many biological pathways, since they are a result of interactions between genomic, epigenomic and environmental variables [ , ]. Therefore, modeling biological pathways across multiple omics domains might help to better understand the underlying genetic architecture and biological processes of complex phenotypes, which in turn leads to improved diagnosis, prognosis and therapy [ ].