NEW ALGORITHMS FOR INTRACTABLE DIRECT METHODS PROBLEMS

该项目确定了一些战略，将用于 将直接定相方法的功率提高到 分子结构比目前确定的常规方式。 除了对一些计算的时间有效的改进之外， 算法中描述的这一建议，见解将提供 关于如何获得更准确和更高分辨率的数据， 可能有利于也可能不利于加强各种分阶段战略。 将记录并检查原生X射线数据，以获得适当的 解决和未知的大分子结构，以测试这些新的 方法. 计算机图形学技术将在未来的计算机科学中发挥重要作用。 加强用于这些大分子的定相程序， 将开发结构和图形软件，以帮助识别和 扩展了存在于边缘相位电子中的结构模式， 密度图 在实验方面，初步研究表明，低温 对单个天然晶体的研究可以提供直接相 以前被认为是不可用的信息。 分析 使用室温和液氮温度数据集， 记录了两种不同的正交短杆菌肽晶体 在两种不同的X射线衍射仪上测量的结构表明， 成功应用这种方法所需的数据精度 并不是不可逾越的 几种疑难结构分析 此外，包括短杆菌肽数据在内的测定结果表明， 可以用最小的努力来确保大的相位块 只有十几个精心挑选和可识别的关键不变量， 可以明确地确定结构解依赖于哪一个。 这些结果是一个令人鼓舞的迹象，复杂的晶体结构， 含有多达300-500个原子， 比目前遇到的结构包含100 原子或更少。

This project identifies a number of strategies that will be used to increase the power of direct phasing methods for applications to larger molecular structures than are currently determined in a routine fashion. In addition to the time efficient improvements to a number of computational algorithms described in this proposal, insights will be provided with regard to how the availability of more accurate and higher resolution data may or may not be advantageous to strengthening various phasing strategies. Native X-ray data will be recorded and examined for an appropriate base of both solved and unknown macromolecular structures to test these new methods. Computer graphics techniques will play an important role in strengthening the phasing procedures used for these large macromolecular structures and graphics software will be developed to help recognize and extend the structural patterns that exist in marginally phased electron density maps. On the experimental side, preliminary studies indicate that low temperature investigations of single native crystals can provide direct phase information that was previously thought to be unavailable. An analysis using a room temperature and a liquid nitrogen temperature data set, recorded from two different crystals of the orthorhombic gramicidin structure, measured on two different X-ray diffractometers, demonstrates that the precision required of the data to successfully apply this method is not insurmountable. Analysis of several difficult structure determinations, including that of the gramicidin data, moreover, shows that large blocks of phases can be secured with a minimum of effort provided that as few as a dozen well-chosen and identifiable pivotal invariants upon which the structure solution depends can be unambiguously determined. These results are an encouraging indication that complex crystal structures containing as many as 300-500 atoms ban be determined with far less difficulty than is currently experienced for structures containing 100 atoms or less.