Applying HMM to find disease-related genes using time-series gene expressiong profiles

• Main Authors: Huang,Yao-Jhen, 黃耀震
• Other Authors: Chang, Weng-Long
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/90361131014278392298

碩士 === 國立高雄應用科技大學 === 資訊工程系 === 102 === Complex disease lack single and affirmative genetic causes, moreover, the prognosis is an important indication of complex disease, such as cancer. To understand the connections between disease and genes, I obtained whole human genome expression data to identify key genes and to assess the association relationship. The time-series data indicated different degrees of similarity and the extent of this discrepancy can be observed and associated with the disease. Initially the data still contain noise information. To retain the useful data for further analysis, I conducted a series of steps to exclude systematic bias and errors. After the data preprocessing, I built a HMM for testing the significant of disease-related genes. The HMM was trained using Viterbi algorithm. My HMM could predict the possible etiology and critical pathogenic genes. In the future, I hope my approach could provide the clinical and research experts the information needed for referencing critical and essential genes for complex disease.