Department of Bioinformatics, School of Life Sciences, Bharathidasan University, Tiruchirappalli 620 024, Tamil Nadu, India.
Department of Biotechnology, Bhupat and Jyoti Mehta School of Biosciences, Indian Institute of Technology Madras, Chennai 600 036, Tamil Nadu, India.
J Struct Biol. 2020 May 1;210(2):107479. doi: 10.1016/j.jsb.2020.107479. Epub 2020 Feb 17.
Annotation of twilight zone protein sequences has been hitherto attempted by predicting the fold of the given sequence. We report here the PredictSuperFam-PSS-3D1D method, which predicts the superfamily for a given twilight zone (TZ) protein sequence. Earlier, we have reported that adding predicted secondary structure information into the threading methods could improve fold prediction especially for the TZ protein sequences. In this study, we have analysed the application of the same method to predict superfamilies. Here, in this method, the twilight zone protein sequence is threaded with the 3D1D profiles of the known protein superfamilies library. In addition, weightage for the predicted secondary structure (PSS) is also employed. The performance of the method is benchmarked with twilight zone sequences. In the benchmarks, 62 and 65 percentages of superfamily predictions are obtained with GOR IV and NPS@ predicted secondary structures, respectively. Receiver Operating Characteristic (ROC) curves indicate that the method is sensitive in predicting the superfamilies. A case study has been conducted with the hypothetical protein sequences of Schistosoma haematobium (Blood Fluke) using this method and the results are analyzed. Our method predicts the superfamily for TZ sequences for which, methods based on sequence similarity alone are inadequate. A web server has been developed for our method and it is available online at http://bioinfo.bdu.ac.in/psfpss.
迄今为止，人们一直试图通过预测给定序列的折叠来注释黄昏带蛋白序列。我们在这里报告了 PredictSuperFam-PSS-3D1D 方法，该方法可预测给定黄昏带（TZ）蛋白序列的超家族。早些时候，我们已经报告说，将预测的二级结构信息添加到线程方法中可以改善折叠预测，特别是对于 TZ 蛋白序列。在这项研究中，我们分析了该方法在预测超家族中的应用。在此方法中，将黄昏带蛋白序列与已知蛋白超家族库的 3D1D 轮廓进行线程处理。此外，还采用了预测二级结构（PSS）的权重。该方法的性能与黄昏带序列进行了基准测试。在基准测试中，使用 GOR IV 和 NPS@预测的二级结构，分别获得了 62%和 65%的超家族预测率。接收者操作特征（ROC）曲线表明该方法在预测超家族方面很敏感。使用该方法对曼氏血吸虫（血吸）的假设蛋白序列进行了案例研究，并对结果进行了分析。我们的方法预测了 TZ 序列的超家族，对于仅基于序列相似性的方法来说，这是不够的。我们已经为我们的方法开发了一个网络服务器，它可以在 http://bioinfo.bdu.ac.in/psfpss 上在线获得。
