Development of statistical methods for integrative omics analysis in precision medicine

Abstract

Precision medicine is an integrative approach to the prevention and treatment of complex diseases such as cardiovascular disease that considers an individual’s lifestyle, clinical information, and omics profile. In the last decade, the advances in omics technologies have allowed researchers to gain insight into biological systems and progress to precision medicine. Many omics technology now enables us to rapidly generate, store and analyse data at a large scale. Many efforts have attempted to integrate large-scale multi-batch and multi-omics data. While many strategies have been developed, challenges remain in developing a robust method cap- able of pre-processing large-scale datasets, handling mislabelled information, and performing integrative analysis. Pre-processing any omics data is essential to remove technical factors whilst preserving biological variance. However, many methods still struggle to mitigate the batch effect, particularly for protracted acquisitions. Furthermore, robust visualisation tools for processing, quality control diagnostics, and integrative analysis of omics data are still lacking in effective data visualisation and integration. Lastly, cell type annotation remains a key challenge in single-cell transcriptomic data analysis due to the incompleteness of our current knowledge and the human subjectivity involved in manual curation. Together, these may result in cell type mislabelling and potentially lead to false discoveries in downstream analysis. This thesis first introduces each of the above challenges in detail (Chapter 1). We then introduce novel strategies and robust methods for the removal of unwanted variation in large-scale metabolomics data (Chapter 2), visualisation tools for omics data diagnostics and integrative analysis (Chapter 3), and cell-type identification methods in single cell transcriptomics data (Chapter 4). Chapter 5 summarises the contributions of each chapter to precision medicine and concludes the thesis.