The Bioinformatics Programme

Genomic studies of Alzheimer’s Disease (AD) have identified high confidence risk loci that serve as a platform for biological investigation leading to novel therapies.

There are a number of major challenges in attempting to translate findings from genome-wide association studies (GWAS) into an appreciation of altered biological function. Our work has provided a clearer picture of the genetic architecture of AD, begun to unmask potentially critical areas of biology, and challenged the utility of existing diagnostic approaches for future research. The challenges now are to implicate individual mutations, genes, and specific biological processes in order to drive mechanistic studies, and to conduct analyses across AD and related disorders to identify novel strata for clinical and interventional studies.

Achieving this will require a programmatic approach and the generation of a considerable body of new omics data, e.g. ChIP-seq, RNA-seq and genomic screens, as well as the integration of all classes of genomic data with robust genetic findings from other disorders and novel sources of multi-omics data. This programme will bridge the gap between statistical genetic association and tractable biological mechanisms and dissect complex pathways using novel mathematical approaches.

Aims

Integrate biological data with large genetic datasets to detect disease-relevant mechanisms

To detect complex patterns in omics data we plan to integrate biological data with large genetic datasets to detect disease-relevant mechanisms, specifically to:

• integrate genetics and biological pathways to refine risk models for the disease
• identify functional variants and true target genes using epigenomic profiling of tissues/cells relevant to innate immunity: microglia, astrocytes, etc.
• use functional genomic approaches to dissect the transcriptional response of disease relevant cell types to environmental exposures.

Apply non-linear multivariate mathematical approaches for disease stratification

To apply state of the art statistical methods and develop novel mathematics, we will bring interdisciplinary experts from mathematics and statistics into AD research. We have already made advances into methodological improvement on traditional polygenic risk scoring by adjusting the SNP effect sizes used in the risk score using spectral decomposition of the SNP correlation matrix (the applicants have established collaboration and publications track record with the School of Mathmatics, Cardiff University).

We will employ data-driven approaches to identify hidden patterns of complex phenotypic/omics data and utilize machine learning approaches to account for nonlinear relationships between prediction variables.

Develop novel mathematical approaches tailored to identifying patterns in non-linear multidimensional “omics” space

Combining public data with bespoke in-house generated data (e.g. single cell omics, epigenetics) will allow us to build epigenomic maps of human microglia and, in the future, of other human brain cell types.

As part of this program we plan to test for interaction with environmental stimuli (e.g. immunogens and stress) and with AD genetics to identify risk loci that operate in microglia. We shall investigate the impact of AD risk variants on the activity of regulatory elements and link AD risk variants to gene targets. This will move genetic risk for AD from statistical associations to defined molecular mechanisms delivering tractable targets and cellular contexts for further modelling.