论文部分内容阅读
合成生物学中标准化元件数量多、种类杂,使得构建生物设备时难以选择标准化元件,将这些元件可视化有助于提高生物设备构建效率.考虑生物元件为长度不一的基因短序列,文章通过结合编辑距离与高斯核函数构建生物元件距离矩阵,使用拉普拉斯特征映射方法将生物元件序列降为二维或三维数据;通过图形化展示,功能类似的生物元件可有效地聚类,功能差异大的元件可有效地区分,且对降维后数据聚类显示的二分类精度达到91.6%,三分类精度达到82.4%.实验结果表明,降维后的数据具有良好的区分度,通过降维可视化将显著提高标准化元件的选择效率.“,”In synthetic biology ,there are a number of standard parts with a wide variety of categories , making it hard to choose a part w hen constructing devices .Visualizing these parts could simplify the part selection .Considering that synthetic biological parts are DNA segments with various lengths ,the similarity of these parts is evaluated by the integration of edit distance and Gaussian kernel .Based on the similarity ,Laplacian Eigenmaps is employed to reduce data dimensions to two or three dimen-sions .By visualizing the reduced data ,the parts with similar functionality could cluster together ,and the parts with different functionality could be separated efficiently .Besides ,the cluster accuracy for two kinds and three kinds of parts reaches 91 .6% and 82 .4% ,respectively ,which proves the discrim-ination of the reduced data ,and this could significantly improve the efficiency of parts selection .