Age is the major risk factor for a variety of non-communicable diseases. As life expectancy increases, ageing poses significant challenges to the individual, society, and healthcare systems. Ageing is a complex process involving multiple interconnected cellular and organismal phenotypes. Thus, understanding the molecular mechanisms and finding potential interventions is challenging and requires systems-level approaches. In this PhD I have addressed three main questions about ageing, using high-throughput data and computational methods.

My first study considers interindividual heterogeneity in gene expression during ageing. Previous studies had suggested that phenotype, epigenome and gene expression become more heterogeneous with age. However, the list of genes and pathways reported as heterogeneous in late age showed differences in the literature and did not resolve whether the increase in heterogeneity is a time-dependent process starting at birth or is restricted to the ageing period (i.e. after 20 years of age). Using different data pre-processing steps and heterogeneity measures on the same transcriptome dataset, we have shown that the inconsistency in the literature could reflect technical issues as well as biological variability. Next, applying a meta-analysis scheme that relies on consistent results across multiple datasets to increase reproducibility, we have shown that the increase in inter-individual heterogeneity starts after the age of 20. Moreover, the genes that become more heterogeneous during ageing have a higher number of transcriptional regulators (miRNAs and transcription factors) and are associated with known longevity pathways.

My second study focuses on the link between ageing and age-related diseases. Many diseases show age-dependency, but the molecular nature of this relationship is not fully understood. Using UK Biobank data, I have characterised 116 common diseases based on their age-of-onset profiles and genetic associations. I first showed diseases following the same age-of-onset distribution are genetically more similar, and this similarity could not be explained by disease categories, co-occurrences, or causal relationships. Two groups of diseases showed age-dependent profiles, starting to become more prevalent after the ages of 20 and 40 respectively. They both showed an association with known ageing-related genes but had different functional and evolutionary profiles. I found support for the two evolutionary genetic theories of ageing, mutation accumulation, and antagonistic pleiotropy, using the variants linked to diseases with different age-of-onsets. I also identified some drugs that could be repurposed to target multiple conditions and potentially decrease the need for polypharmacy in the elderly.

Finally, I followed a systems-level approach to identify drugs that can target ageing in the human brain. Using transcriptome datasets from multiple brain regions, I first identified the gene expression changes that can characterise ageing. Then, compared with the drug-perturbed gene expression profiles in the Connectivity Map, I identified 24 drugs that are significantly associated with the ageing signature. Some of these drugs may function as anti-ageing drugs by reversing the detrimental changes that occur during ageing, others by mimicking the cellular defence mechanisms. The drugs that we identified included a significant number of already identified pro-longevity drugs, indicating that the method can discover de novo drugs that ameliorate ageing.