Integrative Structure Modeling of the Core SPB

大分子组装体的结构建模最好作为计算优化问题来处理，其中 有关组件的信息被编码到用于评估候选模型的评分函数中。至 为此，我们正在开发集成建模平台(9)(http://inteqrativemodelinq.orq/)，和 开源软件包满意构建大分子蛋白质复合体的空间模型 来自各种实验数据的空间约束。IMP通过几个应用实例进行了演示 复合体，包括核孔复合体(4)和26S蛋白酶体(11)。我们将使用IMP Inan 迭代实验和计算过程构建酵母的伪原子分辨模型 主轴极体(SPB)。模型构建的空间约束将从当前可用的 数据以及在拟议的POL中心产生的新数据。我们将首先改进、发展和 使用结晶学、电子显微镜(EM)、小角X射线的基准计算协议 散射(SAXS)、Forster共振能量转移(FRET)光谱和交联型数据 SPB的结构建模。实验数据中的不完备性、噪声和模糊性将被处理 使用将从数据中提取最大可能信息的基于推理的评分函数， 采用最小假设和近似的贝叶斯方法(13)。对所有模型结构进行抽样 与数据保持一致通常是一个具有挑战性的问题，部分原因是评分的崎岖性质 功能景观及其许多局部极小值。为此，我们将开发一种高效的采样方法 将完整的自由度集划分为可能重叠的子集，找到候选解 对于子集，由传统的优化器或枚举器独立进行，然后再组合兼容 解决方案，以获得整个系统的保证最佳得分解决方案(15)。最后，我们将应用这些 打分和取样方法确定SPB的伪原子结构。 小灵通398/

Structural modeling of a macromolecular assembly is best cast as a computational optimization problem where information about the assembly is encoded into the scoring function used to evaluate candidate models. To this end, we are developing IMP, the Integrative Modeling Platform (9) (http://inteqrativemodelinq.orq/), an open source software package to construct spatial models of large macromolecular protein complexes by satisfaction of spatial restraints from a variety of experimental data. IMP was demonstrated by application to several complexes, including the Nuclear Pore Complex (4) and the 26S proteasome (11). We will use IMP inan iterative experimental and computational process to construct a pseudo-atomic resolution model of the Yeast Spindle Pole Body (SPB). The spatial restraints for model construction will be derived from the currently available data as well as new data generated in the proposed POl center. We will first improve, develop, and benchmark computational protocols for using crystallography, electron microscopy (EM), small angle X-ray scattering (SAXS), Forster resonance energy transfer (FRET) spectroscopy, and cross-linking data for integrative structure modeling ofthe SPB. Incompleteness, noise, and ambiguity in experimental data will be treated using an inference-based scoring function that will extract the maximum possible information from the data, following a Bayesian approach with minimal assumptions and approximations (13). Sampling all model structures consistent with the data is generally,a challenging problem, in part due to the rugged nature ofthe scoring function landscape and its many local minima. To this end, we will develop a sampling method that efficiently divides the complete set of degrees of freedom into potentially overiapping subsets, finds candidate' solutions for the subsets independently by traditional optimizers or enumeration, and then combines compatible solutions to obtain guaranteed best-scoring solutions for the whole system (15). Finally, we will apply these scoring and sampling methods to determine the pseudo-atomic structure of the SPB. PHS 398/