A New Approach to Rapid Protein-Protein Docking

大分子复合体构成了生命的机器，与许多疾病的理解有关。 比如癌症和新陈代谢紊乱。对这些结构的了解不仅可以提供机械上的 对这些复合体如何发挥作用的描述，以及开发相关治疗干预的线索 为了疾病。预测这些多组分的生物分子复合体(“蛋白质-蛋白质对接”)是关键 技术在发展这类知识方面的作用。我们提出了一种独特的结构分子之间的合作 生物学家、应用数学家和计算机科学家开发和优化新的算法和 将它们集成到灵活的停靠工作流环境中。我们设想提高速度，效率， 预测、可视化和分析蛋白质相互作用的通用性和灵活性。由此产生的 应对软件进行校准、验证，并向学术界免费提供。 我们对当前方法的分析表明，通过利用和发展最先进的数学 模型和算法可以显著提高蛋白质-蛋白质相互作用的预测。这些 改进解决了当前对接方法中的各种限制，包括：1)适用于 广泛的生物系统：拟议的空间高效、多分辨率、体积表示 分子形状适用于任何蛋白质拓扑结构，它显著增加了能够 可计算性；2)可扩展性：建议的表示法允许我们使用 方法类似于我们的分子形状表示，并可扩展到柔性蛋白质；3)空间和 时间效率：我们的新表示法自然有助于极快的空间效率搜索和 利用一种新的自适应不规则采样傅立叶计算进行评分4)软各向异性对接亲和度 功能：我们对蛋白质相互作用的评分是基于定义的分析对接亲和力函数 关于分子界面体积，允许软对接以及例如建模的能力 水在界面等方面具有生物意义。

Macromolecular complexes form the machinery of life, and are relevant to understanding many diseases such as cancer and metabolic disorders. Knowledge of these structures can provide not only the mechanistic descriptions for how these complexes function but also clues in developing therapeutic interventions related to disease. Prediction of these multi-component biomolecularcomplexes ("protein-protein docking") is a key technology in developing such knowledge. We propose a unique collaborationbetween structural molecular biologists, applied mathematicians and computer scientists to develop and optimize novel algorithms and integrate them into a flexible docking-workflow environment. We envision improving the speed, efficiency, generality and flexibility of predicting, visualizing and analyzing protein-protein interactions. The resulting software shall be calibrated, validated and made freely available to the academic community. Our analysis of current approaches indicates that by utilizing and developing state-of-the-art mathematical models and algorithms we can significantly improve the prediction of protein-protein interactions. These improvements address a variety of limitations in current docking approaches, including: 1) Applicabilityto a wide range of biological systems: The proposed space-efficient, multi-resolution, volumetric representationof molecular shape is usable for any protein topology and it significantly increases the size of systems that can be computed; 2) Extensibility: The proposed representation allows us to capture molecular properties using methods similar to our representation of molecular shape and is extensible to flexible proteins; 3)Space and Time Efficiency: Our novel representation lends itself naturally to extremely rapid space efficient search and scoring utilizing a novel adaptive irregularly sampled Fourier calculation;4) Soft Anisotropic DockingAffinity Functions: Our scoring of protein-protein interactions is based on analyticdocking affinity functions defined on molecular interface volumes allowing for both soft docking as well as the ability, for instance, to model water in the interface and other aspects of biological significance.