癌症藥物活性與基因相關性分析系統

Exploration of the Relationship between Drug Activities and Gene Expression in Diseases

陳信志(Austin H. Chen) ； 謝仁傑(Jen-Chieh Hsieh)

《醫療資訊雜誌》 16卷3期 (2007/09) Pp. 49-60

https://doi.org/10.29829/TJTAMI.200709.0004

摘要

在疾病的治療中，了解藥物與基因之間的相關性是很重要的。在疾病發生初期，使用正確的藥物在疾病上做治療，將有助於治癒重大死亡疾病，並降低日後的風險。隨著藥物的製造技術發展，不同的藥物運用在治療相同疾病，甚至相同藥物也會對其他疾病產生作用；在臨床應用上，醫師必須使用最有效的藥物，才能產生最好的效用。由於藥物活性對疾病有太多的作用組合，導致醫生在臨床應用上容易產生混淆，甚至對於用藥的精確率有所偏差，導致判讀上的不便。本論文中，我們開發了藥物活性與基因表現的分析系統，讓藥物活性和基因表現的資料，經由貝氏網路K2演算法的因果關係分析技術、Pearson相關係數分析技術及分群技術將疾病中相關的基因及治療有正面效用的藥物篩選出來，同時產生各項藥物對疾病的關係圖。臨床醫療人員能夠閱讀圖形化或相關性資料，整合疾病對基因、疾病對藥物、藥物對基因等不同的關係，選擇最有效的藥物來治療。本論文應用電腦工具的輔助，利用電腦技術（程式語言Matlab）的處理來建立使用者圖形介面後，實作各項分析方法。挑選了四種方法作為分析藥物活性對於癌症及其基因表現分析的方法：K2演算法、Pearson相關係數分析技術、K-mean分群技術以及Hierarchical階層分群技術，能隨時針對其特徵（藥物或疾病）做分析。醫療人員只要選擇其所需要的分析技術或相對應的資料，經由分析系統處理，即能藉由程式得到所需的資訊或輸出Microsoft EXCEL(XLS)的檔案，來解決用藥問題。

關鍵字

分析系統 ； 疾病 ； 藥物活性 ； 基因表現分析 ； 貝氏網路

並列摘要

Understanding the relationships between drug activities and genes are important in the treatment of diseases. Using the right medicines to cure the diseases at their early stage will be more effective. With the development of new medicines, the amount of medicines used to cure the same diseases is increasing. Some medicines are also capable of curing different diseases. In clinical practices, doctors need to use the effective medicines for a specific disease. Two datasets are especially useful in the study of the relationships between drug activities and gene expression of cancers. One is the relationships between drug activities and cancers, and the other is gene expression verse cancers. In this paper, we integrated both datasets into a drug activities and gene expression system using four analysis techniques. The goal of this study is to design a system platform to display various relationships among drug activities, gene expression, and cancers. These graphical interfaces are developed by Matlab. The relationships between medicines and gene expression profiles were constructed based on four analysis techniques, including the K2 algorithm, the K-means algorithm, the Pearson correlation analysis technique, and the Hierarchical clustering technique. The system can not only generate the most effective group of medicines for any selected cancers but also can output drugs to genes Microsoft EXCEL (XLS) files to medical professionals.

並列關鍵字

Analysis System ； Disease ； Drug Activity ； Gene Expression ； Bayesian Network