在奈米碳管回歸模型以及細胞毒性分類模型在支持向量機上的應用

Abstract

近年來高通量篩選被廣泛應用在生物醫學及生物資訊領域上，傳統的生物實驗非常的耗費時間以及金錢，而且速度又慢，本篇論文是利用機器學習的方法，針對兩種不同的主題進行探討和研究。 第一個主題是針對jurkat細胞株的高通量篩選的實驗數據，利用支持向量機來建立數值模型，希望能夠得到很好的分類效果，並且幫助生物學家的再做實驗前，能夠有好的工具來篩選那些較有機率是對jurkat細胞株有毒性結構或是沒有毒性的結構，進而提升工作的效率。 第二個主題是奈米碳管活性的研究，近年來奈米碳管常常被應用在生物醫學上，當作生物感測器或是藥物攜帶者，所以奈米碳管對於人體是否有毒性越來越被人們所重視。在本篇論文當中，希望能夠透過一組經由奈米碳管的表面的修飾來降低在人體內所造成的毒性的實驗數據，經由基因演算法，找出針對牛血清白蛋白、碳酸酐酶、胰凝乳蛋白脢、血紅素以及細胞存活率、一氧化氮反應，這六種活性的最佳數值模型，其中牛血清白蛋白、碳酸酐酶、胰凝乳蛋白脢、血紅素是探討蛋白質結合的情況，而細胞存活率、一氧化氮反應，前者是探討細胞存活力，後者是有關人體免疫反應，最後進行模型以及結構上的探討。

In recent years, high-throughput screening is widely used in the field of biomedical and biological information. Traditional biological experiment is very time and money consuming. The thesis is discussing and studying about the use of machine learning methods for two different topics. The first topic is building model for high-throughput screening data of jurkat cell line by support vector machine. The goal is to find out a good classification result and suggest biologists whether certain compounds are active or inactive to jurkat cell line before biological experiments. The second topic in the thesis is the study of carbon nanotubes activity. In recent years, carbon nanotubes are widely used in biomedical field and as biosensors or drug carriers. Therefore, the issue of carbon nanotube toxicity to human body is getting more and more important. In this thesis, we modify the surface of carbon nanotubes to find the relation between attached decorators and the complex toxic activity. There are six types of biological activity to be researched which are bovine serum albumin (BSA), carbonic anhydrase(CA), chymotrypsin(CT), hemoglobin, cell viability and NO response. The first four biological activities are related to protein binding, while cell viability and NO response are related to cell apoptosis and human immune response. Finally, we discuss results between models and the feature of compounds.