Novel method for conditional and interaction gene-set analysis

18 September 2018

PhD student Christiaan de Leeuw (CNCR-CTG) has developed an important new tool for advanced gene-set analyses, published in Nature Communications this week

Genetic discovery studies routinely yield large numbers of associated genes for multiple traits. Gene-set analysis aids in interpreting the combined function of these genes and may provide insight into whether associated genes implicate specific biological pathways and processes, cellular functions, tissue-specific expression or other functions of genes.

However, current gene-set analysis methods cannot account for the multiple number of functions one gene may have. Christiaan de Leeuw from CNCR-CTG has now developed a novel method, implemented in the MAGMA software, to address this shortcoming. The new method allows for the simultaneous analysis of any number of gene functions. This can e.g. show how the effects of different functions of a gene relate to each other in the context of the trait under investigation, and can help detect or rule out confounding between the multiple functions. The model also allows for interaction analysis, providing the ability to test scenarios such as tissue-expression dependent involvement of specific biological processes.

We applied this new approach to blood pressure phenotypes in the UK Biobank data. The analysis revealed numerous gene characteristics associated for these phenotypes, but also widespread confounding between them. The model could narrow these results down to a more clearly defined subset of gene functions however, pointing to cardiovascular and smooth muscle cell involvement as well as a role for cell proliferation and intracellular regulation. Tissue-specific expression was also implicated, identifying arterial expression as well expression in the female reproductive organs. Following up on this, new associations for several additional biological processes were discovered that are dependent on expression in these tissues. These results suggest that specifically the subset of genes involved in these processes that are also strongly expressed in arterial or female reproductive tissues play a role in the genetics of the blood pressure phenotypes.

Christiaan A. de Leeuw, Sven Stringer, Ilona A. Dekkers, Tom Heskes & Danielle Posthuma Conditional and interaction gene-set analysis reveals novel functional pathways for blood pressure. Nature Communications 9:3768 (2018)