BindingDB: An Open Knowledgebase of Protein-Small Molecule Interactions

结合特定蛋白质的有机小分子是最有效的方法之一 医生必须治疗疾病，研究人员可以用来探测生命系统。 这样的小分子，也称为配体，可以以许多方式起作用，例如通过阻断一个或多个配体， 通过激活蛋白质，或通过使蛋白质被分解， 正常的细胞过程。事实上，大多数药物都是配体， 大学，政府实验室和制药公司，一直在努力寻找 比如药物和生物探针。这些持续的努力产生了持续的 关于什么小分子结合什么蛋白质以及结合的紧密程度的信息。 此信息不仅在生成它的特定项目中有用， 其他应用，如帮助研究人员识别探针分子，以帮助他们 研究，作为计算化学家创建软件的基准， 预测配体-蛋白质结合，以及训练和测试用于药物设计的机器学习工具。 然而，产生这些信息的科学家通常会在科学文章或专利中发布， 其他研究人员无法轻易找到或访问。 该项目的核心目的是进一步开发BindingDB知识库， 极大地扩展了蛋白质-配体结合信息的可用性， 我们需要将信息与其他知识领域相结合，以使其尽可能广泛地发挥作用。这 将通过使用自动化和人工方法相结合来实现， 从科学文章和专利中快速、准确地提取大量数据。这些数据 将以机器可读的格式呈现，与相关数据链接，例如关于 蛋白质的结构和功能，并通过开放源代码格式公开提供， 可搜索的BindingDB网站，该网站还允许离线下载大量数据 使用. BindingDB中的信息将根据高社区标准进行管理， 可查找性，可访问性，互操作性和可重用性（FAIR），该项目将实现 高CoreTrustSeal标准和可靠性和长期保存认证。在 此外，还将采取措施，最大限度地提高这些信息的可用性和整合性，例如 将其作为新兴云资源中的公共数据集提供，并从 行期刊文章和专利到BindingDB中从中提取的数据。

Small, organic molecules that bind specific proteins represent one of the most effective ways that physicians have to treat diseases and that researchers can use to probe living systems. Such small molecules, also known as ligands, can act in many ways, such as by blocking a protein from working, by activating a protein, or by causing the protein to be broken down by normal cellular processes. In fact, most medications are ligands, and researchers in universities, government labs, and pharmaceutical companies, are constantly at work seeking new ones as drugs and biological probes. These ongoing efforts generate a continuous flow of information about what small molecules bind what proteins, and how tightly. This information is useful not only within the specific project that generated it, but also for many other applications, such as helping researchers identify probe molecules to help with their research, serving as benchmarks for computational chemists creating software designed to predict ligand-protein binding, and training and testing machine-learning tools for drug design. However, scientists generating this information typically release it in scientific articles or patents, where it cannot easily be found or accessed by other researchers. The core purpose of this project is to further develop the BindingDB Knowledgebase, dramatically expanding the availability of protein-ligand binding information and connecting this information to other areas of knowledge in order to make it as broadly useful as possible. This will be accomplished by using a combination of automated and human methods to carry out fast, accurate extraction of large volumes of data from scientific articles and patents. These data will be rendered in machine readable format, linked with related data, such as information on protein structure and function, and made publicly available in open source format via the searchable BindingDB website, which also allows data to be downloaded in quantity for offline use. The information in BindingDB will be managed according to high community standards for findability, accessibility, interoperability, and reusability (FAIR), and the project will achieve the high CoreTrustSeal standards and certification for reliability and long-term preservation. In addition, steps will be taken to maximize usability and integration of this information, such as by making it available as a public dataset in emerging cloud resources and creating links from on- line journal articles and patents to the data extracted from them in BindingDB.