Method Development: Efficient Computer Vision Based Algo

随着实验数据的快速增长和计算方法的最新进步，现代生物学朝着解决最具挑战性的问题之一--蛋白质功能的预测--又近了一步。理解蛋白质最基本的功能需要了解分子间的相互作用。目前，人们普遍认为，用于分析和预测的累积数据的规模需要具有适当应用能力的高效计算工具。我们正在开发用于结构模式发现和分子缔合预测的计算方法。我们侧重于它们在一系列生物学问题上的应用，以及这些方法结合以及它们与生物实验相结合的优势。我们协同地将结构建模、刚性和柔性结构对齐、保守结构模式的检测以及对接(刚性和柔性与铰链弯曲运动)结合在一起。我们的目标是更广泛地利用计算方法，并使它们与实验相互促进。大多数蛋白质在结合成多分子组合时起作用。然而，对多分子络合物结构的预测在很大程度上还没有得到解决，可能是因为这个问题的组合复杂性很大。传统上，对接应用被用来预测分子之间的成对相互作用。我们开发了一种算法，将对接的应用扩展到多分子组装。我们将其应用于预测齐聚物和多蛋白质复合体的四元结构。该算法很好地预测了我们数据集中所有情况下输入亚单位的接近自然的排列，其中不同目标复合体的亚单位的数量从三个到十个不等。为了模拟更真实的场景，测试了未绑定的用例。在这些情况下，亚基的输入构象要么是亚基的未结合构象，要么是通过同源建模技术获得的模型。对亚基的输入构象与其在目标复合体中的构象不同的未结合情况的成功预测表明，该算法是稳健的。我们期望这种类型的算法对于预测大分子组装体的结构特别有用，这些大分子组装体的结构很难通过实验结构确定来解决。与刚性对接相比，灵活对接算法FlexDock的独特之处在于，它能够处理柔性分子中的任意数量的铰链，而不会降低运行时性能。具有相似物理化学性质和形状的蛋白质表面区可能执行相似的功能并结合相似的结合伙伴。我们开发了识别结合位点和界面相似性的算法和软件包。这两种方法都识别局部的几何和物理化学相似性，即使在没有全序列或折叠相似性的情况下也可以存在。第一种方法，SiteEngine，接收两个蛋白质结构作为输入，并搜索一个蛋白质的完整表面，以寻找与另一个蛋白质结合位点相似的区域。第二个，界面到界面(I2I)-SiteEngine，比较了蛋白质-蛋白质界面，这是两个蛋白质分子之间相互作用的区域。它接收蛋白质-蛋白质复合体的两种结构作为输入，提取界面，并找到使两对相互作用的结合位点之间的相似性最大化的三维转换。输出包括PDB文件格式的叠加和被比较的实体共享的物理化学性质的列表。这些方法效率很高，而且免费提供的软件包适用于整个PDB的大规模数据库搜索。

The rapid increase in experimental data along with recent progress in computational methods has brought modern biology a step closer toward solving one of the most challenging problems: prediction of protein function. Comprehension of protein function at its most basic level requires understanding of molecular interactions. Currently, it is becoming universally accepted that the scale of the accumulated data for analysis and for prediction necessitate highly efficient computational tools with appropriate application capabilities. We are developing computational methods for structural pattern discovery and for prediction of molecular associations. We focus on their applications toward a range of biological problems and the advantages of the combination of these methods and their integration with biological experiments. We synergistically merge structural modeling, rigid and flexible structural alignment and detection of conserved structural patterns and docking (rigid and flexible with hinge-bending movements). Our goal is toward a broader utilization of computational methods, and their cross-fertilization with experiment. The majority of proteins function when associated in multimolecular assemblies. Yet, prediction of the structures of multimolecular complexes has largely not been addressed, probably due to the magnitude of the combinatorial complexity of the problem. Docking applications have traditionally been used to predict pairwise interactions between molecules. We have developed an algorithm that extends the application of docking to multimolecular assemblies. We apply it to predict quaternary structures of both oligomers and multi-protein complexes. The algorithm predicted well a near-native arrangement of the input subunits for all cases in our data set, where the number of the subunits of the different target complexes varied from three to ten. In order to simulate a more realistic scenario, unbound cases were tested. In these cases the input conformations of the subunits are either unbound conformations of the subunits or a model obtained by a homology modeling technique. The successful predictions of the unbound cases, where the input conformations of the subunits are different from their conformations within the target complex, suggest that the algorithm is robust. We expect that this type of algorithm should be particularly useful to predict the structures of large macromolecular assemblies, which are difficult to solve by experimental structure determination. The flexible docking algorithm, FlexDock, is unique in its ability to handle any number of hinges in the flexible molecule, without degradation in run-time performance, as compared to rigid docking. Protein surface regions with similar physicochemical properties and shapes may perform similar functions and bind similar binding partners. We developed algorithms and software packages for recognition of the similarity of binding sites and interfaces. Both methods recognize local geometrical and physicochemical similarity, which can be present even in the absence of overall sequence or fold similarity. The first method, SiteEngine, receives as an input two protein structures and searches the complete surface of one protein for regions similar to the binding site of the other. The second, Interface-to-Interface (I2I)-SiteEngine, compares protein-protein interfaces, which are regions of interaction between two protein molecules. It receives as an input two structures of protein-protein complexes, extracts the interfaces and finds the three-dimensional transformation that maximizes the similarity between two pairs of interacting binding sites. The output consists of a superimposition in PDB file format and a list of physicochemical properties shared by the compared entities. The methods are highly efficient and the freely available software packages are suitable for large-scale database searches of the entire PDB.