Chemical Pattern Detection and Visualization in Biological Networks

描述（由申请人提供）：大量的分子生物活性数据创造了新的机会，但由于处理许多独立和异构数据源的固有困难，它也阻碍了科学家进行生物医学研究的方式。数据输入的质量和类型通常会限制项目的结果。为了提高研究成果，获取所有可用的数据和多重可选假设检验是必不可少的。针对经验不足的最终用户，我们将开发工具，促进小分子/生物活性/生物医学数据领域的“跳跃”，从一个潜在的解决方案引导到另一个，鼓励用户探索多个替代假设。我们将整合PubChem， ChemBank, ChEMBL， PDSP和WOMBAT等多个生物活性数据库的数据到一个集中的系统中。我们将开发先进的化学模式识别算法，并提供基于cytoscape的可视化工具，用于全球探索化学模式与生物活动/靶标之间的关系。我们将通过三个具体目标来实现这一目标：为许多异构数据库创建一个简单的统一接口，CARLSBAD（小分子生物活性数据联合注释研究库）；这些数据将协调人类、大鼠和小鼠靶点的多种来源的小分子生物活性数据。2. 开发先进的化学模式检测和注释算法；我们将检测最大重叠集（MOS）和分层支架（HierS），并相应地对CARSLBAD中的化学物质进行注释。3. 开发一个Cytoscape插件，用于可视化和探索化学模式生物活性网络。通过MOS/HierS模式，用户将能够识别目标特定的化学特征（活性和选择性的决定因素）；在缺乏特定信号的情况下，这些模式将作为脱靶和混杂生物活性预测的基本原理。存储独特的靶配体生物活性数据以及化学模式，CARLSBAD将在一个企业平台上设计、实施和维护，供科学界使用。新的Cytoscape插件将与现有的核心组件和插件集成，以多学科的方式跨越化学和生物学。

DESCRIPTION (provided by applicant): While the massive amount of molecular bioactivity data creates new opportunities, it also hinders the way scientists conduct biomedical research due to the inherent difficulty of processing many separate and heterogeneous data sources. The quality and type of data input often limits the project outcome. To improve research outcome, access to all available data and multiple alternative hypothesis testing are essential. Targeting less experienced end-users, we will develop tools that facilitate "jumps" in the small molecule / bioactivity / biomedical data area, leading from one potential solution to another, encouraging users to explore multiple, alternate hypotheses. We will integrate data from multiple bioactivity databases, including PubChem, ChemBank, ChEMBL, PDSP and WOMBAT, into one centralized system. We will develop advanced chemical pattern recognition algorithms and deliver a Cytoscape-based visualization tool for the global exploration of relationships between chemical patterns and biological activities/targets. We will achieve this via three Specific Aims: 1. Create one simple unified interface for many heterogeneous databases, CARLSBAD (Confederated Annotated Research Libraries of Small molecule Biological Activity Data); the data will reconcile small molecule bioactivity data across multiple sources for human, rat and mouse targets. 2. Develop advanced algorithms for chemical pattern detection and annotation; we will detect the Maximum Overlapping Set (MOS) and HierS (hierarchical scaffolds) and annotate chemicals in CARSLBAD accordingly. 3. Develop a Cytoscape plugin for the visualization and exploration of chemical pattern bioactivity networks. Via MOS/HierS patterns, users will be able to identify target specific chemical signatures (determinants for activity and selectivity); in the absence of specific signals, these patterns will serve as rationale for off- target and promiscuous bioactivity prediction. Storing unique target-ligand bioactivity data as well as chemical patterns, CARLSBAD will be designed, implemented and maintained on an enterprise platform for use by the scientific community. The new Cytoscape plugin will integrate with existing core components and plugins to bridge across chemistry and biology in a multi-disciplinary manner. PUBLIC HEALTH RELEVANCE: The proposed research aims to empower the chemistry and biology research community with an innovative, network-based tool for mining vast amounts of chemical and biological data. It will provide an effective and improved way for researchers to evaluate, visualize and explore small molecule bioactivity data in a multi-disciplinary manner, thus leading to improved output in human health research.