Exploring the possibility of alternative splicing as a cancer biomarker

• Main Authors: Hsu, Min-Kung, 徐敏恭
• Other Authors: Lin, Yeong-Shin
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/65g57e

博士 === 國立交通大學 === 生物科技學系 === 106 === Alternative splicing has been shown to play important roles in the development of multiple cancer types. This regulatory mechanism participates in virtually almost all of the hallmarks of cancer, including unlimited growth, anti-apoptosis, angiogenesis, metastasis, and immune evasion. Despite the importance of alternative splicing, transcriptome-wide splicing analyses have remained underrepresented in cancer researches. In this thesis, I explore the possibility of using splicing patterns as a biomarker for normal-tumor tissue differentiation and prognosis. My results indicate that (1) Splicing and exonic expression patterns could be used to differentiate tumor from normal tissues with high accuracies. Furthermore, classification models based on the combined expressions of multiple transcripts or exonic regions perform better than those based on expressions of single transcript/exonic region in tumor-normal differentiation; (2) Alternatively spliced transcripts form network modules that only partially overlap with those derived from gene networks. The expression patterns of some transcript modules differ considerably between tumor and normal tissues; (3) My preliminary results indicate that splicing patterns could be used to distinguish patients with good and poor prognosis. Again, the expressions of multiple transcripts yield better prognosis predictions than those of single transcripts. Collectively, these results suggest that alternative splicing could be further explored for its potential applications in cancer diagnostics and therapeutics.