Enhanced 3D Processing of Spherical and Non-Spherical Virus Structures at High Resolution

普渡大学增强了球形和非球形病毒结构的高分辨率3D处理尽管X射线结晶学和核磁共振光谱的进步使得常规地获得中小型生物大分子的原子分辨率结构成为可能，但在大分子化合物的高分辨率结构研究方面还没有相应的进展。例如，通过低温透射电子显微镜确定病毒的结构通常需要几周或几个月的时间才能完成，产生的电子密度图分辨率相对较低，并且主要限于具有高度对称性的颗粒。如此大的复合体的高分辨率结构对于理解细胞生物学是必不可少的。在国家科学基金的支持下，将开发提高三维图像重建速度的方法，并将该技术推广到低对称性粒子。这些新的计算技术将利用图像采集技术的进步，结合高性能计算和并行计算机算法。将开发一种可扩展的计算解决方案，不仅能够应对现有的成像技术，而且能够应对其在未来十年的预计演变。还将开发确定粒子投影方向和三维重建的并行算法。最后，用于检测电子显微照片中粒子位置的算法和程序将得到改进。所有这些程序都将为廉价的个人计算机集群开发。这些算法的发展将为生物学家提供一种计算快速的方法来获得大分子络合物的高分辨率结构。要在分子水平上更完整地了解生命过程，还需要对成百上千个这样的复合体进行高分辨率分析。该奖项还将支持计算机科学家在生物成像方面的培训。

ABSTRACT9986316Dan C. MarinescuPurdue UniversityEnhanced 3D Processing of Spherical and Non-Spherical Virus Structures at High Resolution Although advances in X-ray crystallography and NMR spectroscopy have made it possible to routinely obtain atomic resolution structures of small to medium sized biological macromolecules, there have not been equivalent advances in high resolution structural studies of large macromolecular complexes. For example, the structure determination of a virus by cryo-transmission-electron microscopy typically takes weeks or months to complete, produces relatively low resolution electron density maps, and has mainly been limited to particles with high symmetry. High resolution structures of such large complexes are essential for understanding cell biology. With support from the National Science Foundation, methods to increase the speed of three dimensional image reconstructions and to generalize the techniques to particles with low symmetry will be developed. These new computational techniques will take advantage of advances in image acquisition combined with high performance computing and parallel computer algorithms. A scaleable computing solution capable of coping not only with existing imaging technology but also with its projected evolution over the next decade will be developed. Parallel algorithms for the determination of the orientation of particle projections and for three-dimensional reconstruction will also be developed. Finally, algorithms and programs for detecting particle positions in electron micrographs will be improved. All of these programs will be developed for inexpensive clusters of personal computers. Development of these algorithms will provide biologists with a computationally fast method to obtain high resolution structures of large macromolecular complexes. A more complete understanding of life processes at the molecular level awaits high resolution analysis of hundreds if not thousands of such complexes. This award will also support the training of computer scientists in biological imaging.