The investigation of fusion transcript expression patterns in healthy tissues and gynaecological cancers

The human transcriptome comprises of diverse and interconnected networks of RNA molecules of various structures which oversee retaining cellular homeostasis. Whilst mRNA is known for its crucial role in protein synthesis, they only make up a minute portion of the transcriptome. Many other forms of RNAs, oftentimes known as non-coding RNAs, are involved in the regulation of gene expression and cellular function. Alterations of non-coding RNAs can have equally substantial impact on the dysregulation of cellular functions which can lead to diseases such as cancer. Fusion RNAs, also known as fusion transcripts or chimeric RNAs, are a form of non-coding RNA. It is defined as a hybrid transcript of independent genes derived from chromosomal rearrangements or splicing mechanisms and are recognised by the breakpoint region formed between partner genes that have fused together. They are a hallmark for some cancers as they are often detected as somatic mutations appear in the initial steps of tumorigenesis. These fusions may affect tyrosine kinase activity, expression of the transcription factor and various non-coding RNA functions involved in cancer development. Additionally, such changes in the RNA structure may retain open reading frames and subsequently produce fusion proteins which may enhance, reduce or eliminate their original cellular function. The identification of chimeric RNAs in cancer can lead to discussions of their potential as biomarkers, with further implications of developing targeted therapy. Various fusion transcripts have been detected in cancers such as leukaemia, lymphoma and lung cancer. However, studies which directly report fusion transcripts in gynaecological cancer are scarce. With the rise of next-generation sequencing (NGS) and third-generation sequencing (TGS) comes a rise of innovative approaches to exploring the intricacies of our genetic makeup. Recently, complex transcriptomic studies that can compare the healthy and cancer transcriptomic profiles have become more viable with the utilisation of NGS RNA sequencing and Oxford Nanopore Technology direct-RNA sequencing. Its high sensitivity is a valuable feature particularly with the studying fusion transcripts due to their low-abundance in all tissues and cancer cell-lines. However, the analytical gap between sequencing output and biological interpretation has created an urgent need for robust computational pipelines capable of accurately identifying and characterising fusion transcripts. This thesis integrates both short- and long-read RNA sequencing to design, implement, and validate a comprehensive bioinformatics workflow for fusion transcript detection. Using healthy human tissue datasets, the first part of the thesis details benchmarking of multiple fusion-calling tools, optimisation of read-mapping and filtering parameters, and construction of a consensus-driven pipeline that increases specificity by minimising false positives arising from read-through transcription and technical artefacts. These methodological developments enabled the reliable identification of previously unreported fusion candidates in healthy tissues and established a reference catalogue for downstream comparisons. The second part of the thesis applies this optimised workflow to RNA-seq and direct RNA datasets derived from healthy female reproductive tissues and various gynaecological cancer samples. This analysis led to the detection and annotation of numerous tissue-specific and cancer-associated fusion transcripts, several of which retained open reading frames and showed potential to encode novel fusion proteins. Comparative expression analyses revealed differential fusion profiles between normal and malignant tissues, suggesting candidate biomarkers and providing insights into cancer-associated splicing dysregulation. Functional annotation analysis further highlighted potential roles for specific fusion RNAs in oncogenic signalling, transcriptional deregulation, and other processes relevant to gynaecological tumour biology. A larger number of fusion transcripts was detected in gynaecological cancer samples in comparison to the healthy tissue samples. From the fusion transcript output, a protein structure prediction was performed with fusions transcripts that contained an open reading frame sequence, followed by the identification of retained domain structures. Additionally, there was a notable difference in the frequency of GC, CC and GG dinucleotides as the fusion breakpoint junction, which may act as a recognition site. Together, this work delivers a validated fusion RNA-detection pipeline and generates the first integrated catalogue of fusion transcripts across healthy and malignant female reproductive tissues. It unveils the importance of quality check of raw sequencing data when working with a multiplex bioinformatic pipeline, and the necessity of utilising multiple fusion transcript detection tools to increase the robustness of the results. This thesis also reveals potential biological implications of the detected fusion transcripts, particularly those which contain a likely functional domain. Overall, these findings contribute valuable methodological and biological insights and provide a foundation for future exploration of fusion RNAs as diagnostic markers or therapeutic targets in gynaecological cancers.