Data mining the BMRB/PDB for correlations useful for NMR

对 BMRB/P​​DB 进行数据挖掘以获取对 NMR 有用的相关性

• 批准号: 6960965
• 负责人: MICHAEL R GRYK
• 金额: $ 7.4万
• 依托单位: UNIVERSITY OF CONNECTICUT SCH OF MED/DNT
• 依托单位国家: 美国
• 财政年份: 2005
• 资助国家: 美国
• 起止时间: 2005-09-30 至 2007-09-29
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6960965
• 关键词: bioinformatics; chemical structure; computer data analysis; information retrieval; macromolecule; nuclear magnetic resonance spectroscopy; nucleic acid structure; protein structure; structural biology

DESCRIPTION (provided by applicant): Solving the three-dimensional (3D) structure of biological macromolecules using NMR spectroscopy has been useful for the development of several currently used therapeutic drugs and will likely play a key role in the development of many future drugs. Structure determination by this approach is essentially a 3-step process requiring resonance assignment (determining which nucleus in the molecule gives rise to which NMR signal), derivation of structural restraints from NMR parameters (such as interatomic distance and torsion angle constraints) and structural modeling of the data. It is known that the precision and usefulness of the final structure increases with the number of restraints used in the structure calculation. As both human analysis time and spectrometer time are limiting resources, it is critical to maximize the efficiency of data collection and analysis for both resonance assignment and restraint determination. The use of empirically determined correlations between NMR observables and structural parameters has been vital to the progress of NMR structure determination and newly discovered correlations are certainly expected to generate similar advances. NMR and x-ray structures are currently being determined at a rate of approximately 5,000 per year. The Protein Databank (PDB) now houses the coordinates of over 27,000 structures and the BioMagResBank (BMRB) contains NMR parameters for over 3,000 macromolecules. These databanks provide an enormous resource for discovering novel correlations between the structural features of molecules and the NMR parameters which reflect them. While the architectures of the PDB and BRMB are extraordinary for the archival and retrieval of this data to the global scientific community, the public databanks do not offer the capability for mining this enormity of data for correlations useful to NMR structure determination. This capability is essential in order to fully exploit the vast amounts of available postgenomic data. We propose to develop such a resource and use it to mine for important correlations hidden within the data which are useful in improving the efficiency and effectiveness of NMR analysis.

描述（由申请人提供）：利用核磁共振波谱技术解决生物大分子的三维（3D）结构已经对几种目前使用的治疗药物的开发非常有用，并且可能在许多未来药物的开发中发挥关键作用。通过这种方法确定结构本质上是一个三步过程，需要共振分配（确定分子中的哪个核产生哪个核磁共振信号），从核磁共振参数（如原子间距离和扭转角约束）推导结构约束，以及数据的结构建模。众所周知，随着结构计算中约束数的增加，最终结构的精度和实用性也随之提高。由于人力分析时间和光谱仪时间都是有限的资源，因此最大限度地提高数据收集和分析的效率对于共振分配和约束确定都是至关重要的。利用经验确定的核磁共振观测值和结构参数之间的相关性对核磁共振结构确定的进展至关重要，新发现的相关性当然有望产生类似的进展。目前正在以每年约5000个的速度确定核磁共振和x射线结构。蛋白质数据库（PDB）现在容纳了超过27000个结构的坐标，BioMagResBank （BMRB）包含了超过3000个大分子的核磁共振参数。这些数据库为发现分子结构特征和反映它们的核磁共振参数之间的新相关性提供了巨大的资源。虽然PDB和BRMB的体系结构在向全球科学界存档和检索这些数据方面非常出色，但公共数据库没有提供挖掘这些庞大数据以获取对NMR结构确定有用的相关性的能力。为了充分利用大量可用的后基因组数据，这种能力是必不可少的。我们建议开发这样一个资源，并利用它来挖掘隐藏在数据中的重要相关性，这些相关性有助于提高核磁共振分析的效率和有效性。