Chronic pain – defined as pain lasting more than 3 months – exerts an enormous personal and economic burden, affecting >30% of people worldwide. People with chronic pain do not respond uniformly to treatments and current pain therapies are suboptimal given the substantial patient-specific differences in drug response. Mechanisms underlying patient-specific drug response remain elusive, with a broad range of genetic and environmental factors making a contribution. Patient-specific factors are understood to be another fundamental contributor and driver of differential drug response among individuals, highlighting the growing importance of deploying precision medicine in managing chronic pain.
TwinsUK is a large cohort of same-sex adult twin pairs aged 18-104 years comprising ~15,000 participants years with detailed, longitudinal health records and omics data available. In addition, ~2500 twin participants have objective measures of pain sensitivity (quantitative sensory testing, QST). The study provides a unique resource to (i) explore novel disease and target hypotheses (ii) genetically validate pain pathways, and (iii) elucidate the pathophysiology of human chronic pain.
Aim of the Project: the student will apply artificial intelligence (AI), in particular machine learning (ML)-based data integration techniques, to measures of genomics and other omics collected longitudinally in TwinsUK participants. The aim is to understand and guide development in precision medicine for chronic pain. Specifically, they would develop and apply ML algorithms for system-level integration of multiomic datasets from TwinsUK to:-
(i) characterize molecular entities (genes, proteins etc) and pathways that drive chronic pain
(ii) allow clinical stratification of pain phenotypes leveraging molecular data and biomarkers
(iii) leverage computational methods to elucidate molecular heterogeneity unique to each individual, and prediction of individual-specific response to pain drugs and therapies
(iv) compare and benchmark their computational approach with other state-of-the-art ML approaches
(v) devise replication and confirmation studies using publicly available datasets such as UK Biobank (UKB).
Techniques and Skills Learnt: genomics, multiomics, machine learning, computational biology
Overarching yearly objectives:
(i) Y1: Gain complete understanding of existing TwinsUK data sets and previous research; develop ML approaches for multimodal high dimensional omic datasets
(ii) Y2: Implement ML techniques with a focus on pain phenotypes, datasets and assessments eg. QST to generate hypotheses relevant to chronic pain
(iii) Y3: Compare and benchmark hypotheses against complementary datasets (eg. UKB) and developed computational methods
(iv) Y4: Publish in high impact journal(s) and write up thesis for submission.
