He Bifang, Chen Heng, Huang Jian
School of Medicine, Guizhou University, Guiyang, Guizhou, China.
Center for Informational Biology, University of Electronic Science and Technology of China, Chengdu, Sichuan, China.
PeerJ. 2019 Jun 17;7:e7131. doi: 10.7717/peerj.7131. eCollection 2019.
Selection from phage display libraries empowers isolation of high-affinity ligands for various targets. However, this method also identifies propagation-related target-unrelated peptides (PrTUPs). These false positive hits appear because of their amplification advantages. In this report, we present PhD7Faster 2.0 for predicting fast-propagating clones from the Ph.D.-7 phage display library, which was developed based on the support vector machine. Feature selection was performed against PseAAC and tripeptide composition using the incremental feature selection method. Ten-fold cross-validation results show that PhD7Faster 2.0 succeeds a decent performance with the accuracy of 81.84%, the Matthews correlation coefficient of 0.64 and the area under the ROC curve of 0.90. The permutation test with 1,000 shuffles resulted in < 0.001. We implemented PhD7Faster 2.0 into a publicly accessible web tool (http://i.uestc.edu.cn/sarotup3/cgi-bin/PhD7Faster.pl) and constructed standalone graphical user interface and command-line versions for different systems. The standalone PhD7Faster 2.0 is able to detect PrTUPs within small datasets as well as large-scale datasets. This makes PhD7Faster 2.0 an enhanced and powerful tool for scanning and reporting faster-growing clones from the Ph.D.-7 phage display library.
从噬菌体展示文库中筛选能够分离出针对各种靶标的高亲和力配体。然而，这种方法也会识别出与增殖相关但与靶标无关的肽段（PrTUPs）。这些假阳性结果的出现是由于它们具有扩增优势。在本报告中，我们展示了PhD7Faster 2.0，用于从基于支持向量机开发的Ph.D.-7噬菌体展示文库中预测快速增殖的克隆。使用增量特征选择方法针对伪氨基酸组成（PseAAC）和三肽组成进行特征选择。十折交叉验证结果表明，PhD7Faster 2.0具有良好的性能，准确率为81.84%，马修斯相关系数为0.64，ROC曲线下面积为0.90。1000次洗牌的置换检验结果小于0.001。我们将PhD7Faster 2.0实现为一个可公开访问的网络工具（http://i.uestc.edu.cn/sarotup3/cgi-bin/PhD7Faster.pl），并为不同系统构建了独立的图形用户界面和命令行版本。独立的PhD7Faster 2.0能够在小数据集和大规模数据集中检测PrTUPs。这使得PhD7Faster 2.0成为一种增强且强大的工具，用于扫描和报告来自Ph.D.-7噬菌体展示文库中生长更快的克隆。
