Knapp Bernhard, Omasits Ulrich, Schreiner Wolfgang
Unit for Medical Statistics and Informatics, Section for Biomedical Computer Simulation and Bioinformatics, Medical University of Vienna, A-1090 Vienna, Austria.
Protein Sci. 2008 Jun;17(6):977-82. doi: 10.1110/ps.073402508. Epub 2008 Apr 23.
The prediction of T-cell epitopes is an essential part in virtual immunology. Apart from sequence-based techniques, which achieve good results but fail to give insight into the binding behavior of a certain peptide binding to a major histocompatibility complex, structure-based approaches are another important technique. An essential step is the correct placement of the side chains for a given peptide in cases where no experimental data for the structure are available. To our knowledge, no benchmark for side chain substitution in the area of HLA has been reported in the literature. Here, we present a comparison of five different tools (SCWRL, SCATD, SPDBV, SCit, IRECS) applicable for side chain substitution. Each tool is tested on 29 different HLA-A2 structures with experimentally known side chain positions. Parts of the benchmark are correctness, reliability, runtime, and usability. For validation, the root mean square deviations between X-ray structures and predicted structures are used. All tools show different strengths and weaknesses.
T细胞表位的预测是虚拟免疫学的重要组成部分。除了基于序列的技术（该技术虽能取得良好结果，但无法深入了解特定肽与主要组织相容性复合体的结合行为）外，基于结构的方法是另一项重要技术。在没有该结构的实验数据的情况下，一个关键步骤是为给定肽正确放置侧链。据我们所知，文献中尚未报道HLA领域侧链取代的基准。在此，我们对适用于侧链取代的五种不同工具（SCWRL、SCATD、SPDBV、SCit、IRECS）进行了比较。每个工具都在29种具有实验已知侧链位置的不同HLA - A2结构上进行了测试。基准的部分内容包括正确性、可靠性、运行时间和可用性。为进行验证，使用了X射线结构与预测结构之间的均方根偏差。所有工具都显示出不同的优点和缺点。
