蛋白質藥物目前是國際醫藥市場上最重要的藥物種類之一，在蛋白質藥物中，抗體 藥物是很有產值的一種，預期抗體藥物將會成為未來疾病治療的明日之星。但在過去全 球30 幾年所研究出來的許多抗發炎、抗腫瘤的抗體，但仍是鼠原性抗體，故無法直接 用於人體上進行治療，在過去報告中也指出如何將抗體人類化，但是往往造成人類化抗 體失去功能或是親和力下降的問題。目前雖然有美國Medarex 與日本麒麟啤酒公司共同 以其獨特的技術所開發的基因轉殖小鼠XenoMouse (帶人類抗體基因之老鼠)來製造治療 性人類抗體，但XenoMouse 已被申請為專利，無法廣泛利用，且XenoMouse 是否能製造 與先前具療效鼠源抗體一模一樣的人類抗體，仍是很大的問題。本計畫將以人類免疫缺 失病毒為研究對象，以分子動力學模擬及機器學習的方式，進行抗體藥物的電腦輔助設 計研究，預計分5 個步驟進行，包括：(1)透過序列比對將抗體序列進行人類化；(2)抗原 抗體分子接合；(3)抗原-抗體結合：熱點分析計算；(4)抗體結構、抗原抗體重要結合點 與預測活性分析；(5)電腦輔助預測抗體結構與活性技術平台開發。本計劃研究成果將可 發表至國際期刊，提供相關產業參考，並將結果回饋給學界。 Protein-based drugs are one of the most important types of medication in the world’s drug markets. Amongst the protein drugs, antibody medications are one of the most profitable and most promising therapeutic mainstays of future diseases. In the past 30 decades, many of the developed antibodies for anti-inflammation and anti-tumor purposes were still murine (mouse) based, and cannot be directly used for therapeutic purposes on humans. Although there had been past studies that explored the possibility of humanizing these antibodies, they were hindered by problems such as loss-of-functions and decreased antibody titers. XenoMouse is a strain of transgenic mice, developed by Medarex USA and Japan’s Kirin beverage company, which carries human antibody genes and can produce humanized antibodies for therapeutic purposes. However, XenoMouse technology is patented and cannot be widely used, and there remains the question whether XenoMouse is able to produce human antibodies that are comparable in therapeutic efficacy to their murine counterparts. The current study will focus on the computer assisted design and investigation of antibody-based drugs, using human immunodeficiency virus as the target and combining machine learning and simulations of molecular dynamics. The study will be divided into 5 steps: 1) humanizing the antibody sequence through sequence comparisons, 2) molecular binding of antibody and antigen, 3) hot spot analysis and calculation of antigen-antibody binding, 4) antibody-antigen binding sites, structural and activity predictions, and 5) computer assisted development of technology for predicting antibody structure and bioreactivity. The results from this study may be used for manufacturing references by related industry and feed back to the academic communities.