A Web Service for Fragment-based Selectivity Analysis of Drug Leads

摘要 意义：这项提案的目标是为药物研究人员提供引人注目的新工具，以解决 药物合理设计中的靶外选择性。为此，我们建议使用我们的大型存储库 化学碎片与1,000‘S蛋白质紧密结合的位置图(博尔兹曼图) 跨蛋白质家族片段的空间结合模式的相似性。这将使药物化学家在 攻击在药物发现中困扰临床候选人的非靶标毒性挑战。 我们的团队目前得到了NIH第二阶段SBIR拨款(2R44GM109549)的支持，以计算出 用于基于片段的药物设计的大规模BMAP(1,000‘S的片段图和1,000’S的蛋白质图) Www.boltzmannmaps.com)。在此基础上，需要构建用于选择性组件设计的快速工具。 在比较大量蛋白质的结合模式时，搜索这些片段映射的磅。 通过网络将这一能力交到整个行业的药物化学家手中，有可能 显著改善那些受到非目标毒性负面影响的临床结果，并加快交付速度 给病人提供新药。 创新：比较蛋白质内大量不同同工酶的化学片段结合 家族，基于我们独特的片段结合图库，是一种新的科学方法 选择性设计。为了高效地查询绑定模式的大型存储库，我们需要设计一个新的 几何搜索算法。 目的1：提出一种新的基于几何散列的片段结合模式比较算法。 添加了其他参数，如来自化学势排名的相对自由能等 局部结合部位的物理化学性质。我们将提供基于Web的界面，用于请求和 可视化搜索结果。 目标2：在概念验证项目中验证根据目标1开发的工具，以评估业主的选择性-- 来自sirtuin蛋白家族的127种化合物证实了实验数据。 总体影响：总而言之，拥有一系列不同的跨蛋白质内同工酶的片段结合数据 为解决药物发现中的选择性问题提供了一个独特的机会。通过比较碎片中的- 对大量蛋白质的相互作用模式，以前从未可行过，差异结合的评估- 在蛋白质之间进行调节和不受影响的蛋白质之间的相互作用现在更实用。随着被广泛采用， 该服务将提供一个关键工具，以减少由于偏离目标的相互作用而产生的毒性，从而改善 临床试验成功率。

Abstract Significance: The goal of this proposal is to provide drug researchers with compelling new tools that address off-target selectivity in the rational design of drugs. To do this, we propose to employ our large repository of maps of where chemical fragments tightly bind to 1,000’s of proteins (Boltzmann maps) in searches for differ- ences in spatial binding patterns of fragments across protein families. This will empower medicinal chemists in attacking the off-target toxicity challenge that plagues clinical candidates in drug discovery. Our team is currently supported by an NIH Phase II SBIR grant (2R44GM109549) to computationally produce BMaps on a large scale (1,000’s of fragment maps on 1,000’s of proteins) for fragment-based drug design (see www.boltzmannmaps.com). Building upon this, the need is to build fast tools for the design of selective com- pounds that search these fragment maps in comparing binding patterns across a large number of proteins. Putting this capability in the hands of medicinal chemists across the industry via the Web has the potential to significantly improve those clinical outcomes negatively impacted by off-target toxicities and speed the delivery of new medicines to patients. Innovation: Comparing chemical fragment binding across a large number of diverse isozymes within protein families, based on our unique repository of fragment binding maps, is a new scientific approach to the rational design of selectivity. To efficiently query our large repository for binding patterns requires that we devise a new geometric search algorithm. Aim 1: Develop a novel algorithm for comparing fragment binding patterns using geometric hashing, supple- mented with other parameters such as the relative free energy from the chemical potential ranking and other physiochemical properties of the local binding site. We will provide a Web-based interface for requesting and visualizing search results. Aim 2: Validate the tools developed under Aim 1 in a proof-of-concept project to assess selectivity of a proprie- tary set of 127 compounds across the sirtuin family of proteins confirming experimental data. Overall Impact: In summary, having a body of diverse fragment binding data across isozymes within protein presents a unique opportunity to attack the selectivity problem in drug discovery. By comparing fragment in- teraction patterns across a large number of proteins, never feasible before, an assessment of differential bind- ing between proteins to modulate and proteins to leave unaffected is now more practical. As widely adopted, the service would provide a key tool to reduce toxicities due to off-target interactions, resulting in improved success rates of clinical trials.