Algorithms for the Building of Model Proteins Using Structural Homology

建议进一步开发一种交互式软件产品，以帮助建立蛋白质模型。李博士建议在该程序中改进、整合和测试四种新算法。第一种是将二级结构信息融入到蛋白质保守区自动定位算法中。它将防止不适当的参考蛋白质部分被归类为保守的。第二种是增加了一种在可变区域中生成循环的方法。它们在本质上将比目前在结构数据库中发现的那些更具随机性，并且将跨越更多的构象空间。第三个是系统地搜索可能的侧链构象的算法。这将有助于找到一套自洽的、包装良好的旋转异构体。第四个是将低分辨率晶体数据合并到模型中的算法。它使用模拟退火法，以x-碳坐标为距离约束，在输入的结构信息较少的情况下寻找折叠的蛋白质构象。作为其他模型构建方法的辅助工具，它也可以在更有限的范围内应用。综上所述，这些新方法将对使用该软件产品的研究人员大有裨益。

Further development of an interactive software product to aid in the building of model proteins is proposed. Dr. Lee proposes to refine, incorporate, and test four new algorithms in the program. The first is the incorporation of secondary structure information into the algorithm for the automatic locations of conserved regions in the proteins. It will prevent inappropriate sections of the reference proteins from being classified as being conserved. The second is the adding of a method for generating loops in the variable regions. They will be more random in nature than those currently found in a structural database, and will span more of conformational space. The third is an algorithm for systematically searching possible conformations for side chains. It will help find a set of self-consistent rotamers that pack well. The fourth is an algorithm to incorporate low resolution crystal data into the model. It uses simulated annealing using x-carbon coordinates as distance constraints to find a folded protein conformation with little input structural information. It can also be applied in a more limited way as an adjunct to other model building methods. Taken together, these new methods will be of great use to the researchers who use the software product.