Computational analysis of ncRNA involved in cancer development and memory formation

Abstract

The development of large-scale transcriptomic technologies has challenged many assumptions about the genome. One of the most striking observations is that the majority of the genome is expressed as transcripts that lack protein-coding potential. Many such transcripts are expressed in low quantities and are often only transcribed in certain tissues or cell subpopulations. Transcripts lacking coding-potential over 200 nt have been designated as long noncoding RNAs (lncRNA), many of which are in proximity to protein-coding genes, parading a previously undetected loci-complexity. Subsequent studies investigating individual lncRNAs have revealed that hundreds exhibit regulatory functions that are involved in almost all cellular processes. These transcripts’ characteristic low and specific expression has meant that studying and understanding the role of lncRNAs in health and disease is difficult. Consequently, much remains unknown regarding their expression and functions in different systems. This thesis uses bioinformatics analyses of RNA-sequencing to investigate the transcriptome in esophageal adenocarcinoma (EAC) development and long-term memory formation. Observing multiple developmental stages of cancer, and multiple time-points in memory formation, the investigations described in this thesis aim to elucidate and assign potential function and context for lncRNAs through expression correlation, network analysis, and guilt-by-association analysis. In EAC, this thesis identifies novel dysregulated lncRNAs highly upregulated in cancer. Many upregulated lncRNAs correlate with their neighbouring protein- coding gene. Further analysis suggests lncRNA involvement in the cell cycle and in immunological processes. Furthermore, machine learning reveals a novel 4-gene signature capable of segregating EAC from its precursor tissues on the mRNA level. Using in vivo long-term potentiation (LTP) as a surrogate for memory formation, this thesis identifies 71 novel lncRNAs correlating with known memory-associated genes. Additionally, this analysis observes activation of repeat elements in LTP, suggesting a role for retrotransposition in memory formation. Furthermore, investigation of the promoter epigenome of 954 memory-associated mRNAs, reveals widespread promoter methylation changes correlating with gene expression alterations. In summary, this thesis integrates the analysis of the protein-coding transcriptome with its noncoding counterpart to further characterise and explore the cellular alterations observed in the transition from the homeostasis of health to the genetic disorder of disease.