Gniewek Pawel, Kolinski Andrzej, Gront Dominik
Faculty of Chemistry, Warsaw University, Warsaw, Poland.
J Comput Biol. 2012 Jul;19(7):879-86. doi: 10.1089/cmb.2011.0307. Epub 2012 Jun 25.
The development of automatic approaches for the comparison of protein sequences has become increasingly important. Methods that compare profiles allow for the use of information about whole protein families, resulting in more sensitive and accurate detection of distantly related sequences. In this contribution, we describe a thorough optimization and tests of a profile-to-profile alignment method. A number of different scoring schemes has been implemented and compared on the basis of their ability to identify a template protein from the same SCOP family as a query. In addition to sequence profiles, secondary structure profiles were used to increase the rate of successful detection. Our results show that a properly tuned one-dimensional threading method can recognize a correct template from the same SCOP family nearly as well as structural alignment. Our benchmark set, which might be useful in other similar studies, as well as the fold-recognition software we developed may be downloaded (www.bioshell.pl/profile-alignments).
蛋白质序列比较自动化方法的发展变得越来越重要。比较轮廓的方法允许使用关于整个蛋白质家族的信息，从而更灵敏、准确地检测远缘相关序列。在本论文中，我们描述了一种轮廓到轮廓比对方法的全面优化和测试。基于从与查询序列相同的SCOP家族中识别模板蛋白的能力，我们实现并比较了多种不同的评分方案。除了序列轮廓外，还使用二级结构轮廓来提高成功检测的比率。我们的结果表明，经过适当调整的一维穿线法几乎能像结构比对一样，从相同的SCOP家族中识别出正确的模板。我们的基准数据集可能对其他类似研究有用，我们开发的折叠识别软件也可以从（www.bioshell.pl/profile-alignments）下载。
