Gene fusions and microRNAs in cancer

The genome of cancer cells is unstable. Flaws in the DNA repair mechanisms of these cells can lead to the creation of gene fusions, where parts of two different genes are erroneously combined. Additionally, the expression of microRNAs (miRNAs), small regulatory molecules that control gene activity, is often dysregulated in cancer. In this thesis, we investigate miRNAs, gene fusions, and the interplay between the two in cancer using bioinformatic approaches. We found that miRNA host genes are common in gene fusions and may provide an alternative mechanism to dysregulate their expression. Since gene fusion detection methods are prone to errors, we developed a method to validate fusion transcripts at the genomic level using matched whole-genome sequencing data. Utilizing information on validated fusion events from 910 tumors in The Cancer Genome Atlas, we trained a machine learning classifier to predict which fusion event are real, and demonstrated that this approach can improve the quality of fusion detection. Finally, we investigated the function of the ERBB2-encoded mir-4728 in breast cancer at the transcriptional and translational level, and found that it impacts the level of aromatase and other genes involved in estrogen biosynthesis. These
findings contribute to a growing understanding of the complex nature of the cancer
genome. The papers in this thesis lay a groundwork for further exploration of the
multifaceted roles of both miRNAs and gene fusions in cancer, underscoring the
importance of continued investigation into their roles in cancer initiation, progression, and therapeutic response.