Context-specific mapping of genetic variants to function

Genetic variation plays a critical role in defining human phenotypes and is a major mechanism by which evolutionary selection pressures have shaped human fitness over time. Despite this, the impact of inter-individual genetic variation on key areas of human physiology, such as immune responses, remains poorly understood - not least because identifying the genes that are directly affected by regulatory variants has proven challenging. It is now widely accepted that most human traits and phenotypes arise, at least in part, as a result of variation in non-coding regulatory sequence. Indeed, GWAS have shown that >90% of trait associated variants do not lie in coding regions, but rather in non-coding regulatory elements.

Our understanding of gene expression regulatory mechanisms indicates that these regulatory elements impact gene expression in a highly context specific manner, which depends on both cell type and activation state - for example, following exposure to cytokines or other signalling molecules. However, a detailed understanding of the genetic loci that are important for regulating cell type-specific responses to physiological stimuli remains elusive.

The rise of high-throughput molecular techniques has now made it possible to interrogate the function of genetic variants in specific loci at scale, but until these approaches can be applied in a truly genome-wide manner (e.g. studying every variant in the genome), some form of prioritisation is required. To date, prioritisation methods have often been biased towards variants located in chromatin regions that are highlighted as active in at least one of the very few cell types and states that have existing chromatin activity datasets. This presents a significant limitation in the field and calls for a new approach to more broadly identify physiologically relevant cell states. Furthermore, absence of a comprehensively validated functional variant dataset is presenting an impediment to the development of predictive computational models of variant function and target gene mapping.

This proposal is in direct response to these unmet needs, and aims to address three key gaps in mapping variants to genes and function:

1) Identification of novel trait-relevant cell states under which specific loci are hypothesised to exert effects on gene expression using in silico prediction and leveraging pathway and network analysis methods.
2) Systematic characterisation of functional regulatory loci (using CRISPR/Cas9 technologies) and prioritization of individual functional variants (using MPRA - massively parallel reporter assays) in context specific conditions.
3) Leveraging generated data to build an AI/ML model predicting the transcriptional and physiological impact of genetic variation in a relevant immune cell type and state (identified in step 1).

Finally, we propose to focus on selected variants and phenotypes and to validate our findings using a genotype-selectable NIHR Bioresource.