A Study on New Algorithm for Detecting Optimal Composition of Protein Discontinuous Fragments Alignment

• Main Authors: Tzu-Wei-Yen, 顏子維
• Other Authors: 董其樺
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/52057191796227029503

碩士 === 中華大學 === 生物資訊學系 === 105 === In the field of bioinformatics, protein structural alignment is an area investigated by many researchers. Among the studies, not many algorithms have focused on the superposition alignment of the functional fragment group in the local structural alignment between a pair of proteins. The conventional method of alignment involves a superposition alignment of the overall structure of proteins. As a result, one can only observe the similarities in the overall structure of proteins, but cannot compare the similarities in the local structure of proteins. To resolve this issue, this study proposed a new algorithm, which targets the local structure of the functional protein and transforms it into 3D coordinates. It then performs the calculation and assessment of the protein structural fragment to identify the similarities in the local structure of the proteins and to estimate if these two proteins have the same functionality. During the process of calculation, a structure superposition alignment was first used to perform rotation, panning, and super positioning of the two protein fragment groups, and RMSD was used to calculate each coordinate in order to complete the graph. A minimum-spanning-tree algorithm was then used to determine whether the coordinates are suitable, and the unsuitable coordinates are deleted. At the end, the pair with the most similar structures in the fragment group was exported. The results of this study showed that the average accuracy of our testing data can be as high as 94.2%, and can be presented using a fully automated process. Users can easily identify similar fragments in local structures. In the future, we hope to conduct further studies on applications including identifying new effects or side effects of existing drugs. By identifying protein blocks that are different in terms of the overall structure but similar in term of functionalities, we can estimate whether the proteins can be treated using a known drug or we can investigate whether the drugs cause side effects to the proteins.