Engineering genomic features to better understand the molecular basis of immune mediated diseases and their treatment

Genetic and transcriptional studies have demonstrated that shared aetiology between immune-mediated diseases (IMD) is reflected in shared patterns in both data types, and suggested new targets for treatment.

However, the huge number of variants and genes measured mean that only a minority of potential information in these data has been harnessed, and disease prognosis and treatment success remains variable and unpredictable.

My goal is to overcome this dimensional challenge by developing genomic feature engineering which exploits these shared patterns, to extract new insight from jointly analysing over a hundred existing datasets.

I will generate summary features by tailoring dimension-reduction strategies and applying them to genetic and transcriptomic data from patients and cohorts with related traits measured.

I will investigate how each feature contributes to rare and common IMD risk, and prognostic variability within diseases.

I will correlate features with molecular measurements and clinical data to understand the gene products they represent, and the situations (cell type, disease state/subtype) in which they are relevant.

Finally, through predictive modeling, I will explore the expected impact of targeting these gene products in different diseases and subtypes, to generate, and test, hypotheses about which targets might modify specific IMD activity or progression.