在台灣肝癌是癌症死亡率中最主要死亡疾病。分析基因調控機制可以幫助我們了解肝癌。基因調控分析是利用高通量實驗方法如微陣列晶片和系統生物來分析。雖然微陣列晶片技術可以產生全基因的基因表現量資料，卻仍然受限於樣本數量和晶片成本。統計方法可以挑選在正常細胞和癌細胞中有差異表現量的基因，但仍受限於晶片數量。此外分析共同調控的轉錄因子也是必要的。 在這個研究裡，我們嘗試結合多種微陣列晶片及利用統計方法整合晶片。首先，我們挑選相似的肝癌微陣列晶片。再來，我們嘗試組合不同的實驗室晶片並建立統計模型。最後，我們嘗試分析轉錄調控因子找尋在所有實驗裡可能和肝癌相關的轉錄因子。雖然整合晶片會包含大量的雜訊，但也會增加晶片的數量。如果晶片的數量夠大，對調控分析的基因仍然會好。我們開發一個系統可以結合不同的晶片及分析在基因群上的共同調控轉錄因子。結果證明，我們的基因表單與目前已知的肝癌相關基因是一致的。也證明還有很多和肝癌調控相關的基因和轉錄調控因子是可以被驗證的。 Liver cancer is the mostly death disease of cancer mortality in Taiwan. Analysis of gene regulation mechanism can help us understand in liver cancer. State of art gene regulation analysis uses high-throughput experiment method such as Microarray and system biology in analysis. Although microarray technology can generate gene expressions data of whole genome, the analysis still limited in sample number and array cost. Statistics can select genes that express differently between normal tissue and tumor tissue but will still limited in number of chips. Furthermore, analysis of common transcription factors regulated is also needed. In this work, we try to integrate multiple microarray chips and use statistic methods on the integrated data. First, we select similar liver cancer microarray experiments. Next, we try to integrate different array serious and establish statistic model. Last, we try to do transcription factor analysis to find possible TF related to liver cancer across all array serious selected. Though the integration will include huge amount of noise, it will also increase the chip number. If the chip number large enough, the gene list will still be good enough for regulation analysis. We develop a computer system which can join multiple chips and analysis the common transcription factors on those genes. Result shows that our gene list consistent with current known cancer related genes. Our result also shows lots of possible liver cancer related gene and transcription factors which can be later be verified.
