Structural Bioinformatics of Proteins and Protein Complexes and Applications to Cancer Biology

蛋白质和蛋白质复合物的结构生物信息学及其在癌症生物学中的应用

• 批准号: 10623840
• 负责人: ROLAND L DUNBRACK
• 金额: $ 74.45万
• 依托单位: RESEARCH INST OF FOX CHASE CAN CTR
• 依托单位国家: 美国
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-04-01 至 2028-03-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10623840
• 关键词: Advanced Malignant Neoplasm; Antibodies; Bioinformatics; C-terminal; Cancer Biology; Classification; Clinical; Complex; Computing Methodologies; Cryoelectron Microscopy; Crystallography; Databases; Development; Fox Chase Cancer Center; Homo; Human; Laboratories; Malignant Neoplasms; Methods; Molecular; Phosphotransferases; Play; Protein Dynamics; Protein Family; Proteins; Research; Role; Sequence Alignment; Spectrum Analysis; Statistical Data Interpretation; Statistical Study; Structure; T-Cell Receptor; Tail; Techniques; Technology; Therapeutic; Therapeutic Uses; Universities; Variant; Work; anticancer research; deep field survey; deep learning; design; medical schools; programs; protein complex; protein function; protein structure; protein structure prediction; simulation; structural biology; three dimensional structure; unsupervised learning

Project Summary/Abstract Structural biology has a fundamental role to play in the advancement of cancer biology and the development of cancer therapeutics. With the rapid developments in experimental structural determination (both crystallography and cryo-EM spectroscopy), structure prediction methods (primarily AlphaFold2 and RosettaFold), and molecular simulation methods, we are poised to bring new levels structural information to cancer research. In this project, we will analyze the structural variation and dynamics of protein families commonly associated with cancer development or targets of cancer therapeutics using existing clustering methods for protein loops and new unsupervised learning techniques from the field of deep learning. We will develop methods for using AlphaFold2 to predict the structures of active and inactive kinases using templates based on our classification of active and inactive states of kinases and multiple sequence alignments optimized for this task. In relevant cases, these structure predictions will include the N and C terminal tails and other domains which may interact with the kinase domains. We will integrate AlphaFold2 structure predictions of protein homo- and heterooligomeric complexes with our database of common interfaces and assemblies found across the structures of proteins in the PDB. Interactions observed in crystals that are replicated by AlphaFold2 present well-founded hypotheses for functional protein interactions. This will be applied specifically for all human kinases where homodimer interactions play an important role in activation and inhibition. We will continue our structural bioinformatics studies of antibodies and expand this work to T-cell receptors, and investigate the utility of deep learning methods for computational antibody and TCR design. Finally, we will bring new structure prediction technologies and our statistical analysis of protein structures to the ongoing research programs of laboratory and clinical colleagues at Fox Chase Cancer Center and Temple University School of Medicine.

项目摘要/摘要 结构生物学在癌症生物学的进步和癌症的发展中起着基础性的作用。 癌症治疗学。随着实验结构测定的快速发展(既有结晶学 和低温EM光谱)、结构预测方法(主要是AlphaFold2和RosettaFold)和分子 模拟方法，我们准备为癌症研究带来新的层次的结构信息。在这个项目中， 我们将分析通常与癌症相关的蛋白质家族的结构变化和动态。 利用蛋白质环的现有聚类法和新的方法开发癌症治疗的靶点 来自深度学习领域的无监督学习技术。我们将开发使用AlphaFold2的方法 使用基于我们对活性和非活性的分类的模板来预测活性和非活性的激酶的结构 为这项任务优化的激活酶和多个序列比对的非激活状态。在相关案件中，这些 结构预测将包括N和C末端以及其他可能与该激酶相互作用的结构域 域名。我们将整合蛋白质同源和异源低聚复合体的AlphaFold2结构预测 利用我们的数据库，在PDB的蛋白质结构中发现了共同的接口和组装。 在由AlphaFold2复制的晶体中观察到的相互作用提出了有充分依据的假设 功能性蛋白质相互作用。这将专门应用于所有人类同源二聚体 相互作用在激活和抑制中起着重要作用。我们将继续我们的结构生物信息学 研究抗体并将这项工作扩展到T细胞受体，并调查深度学习方法的实用性 用于计算抗体和TCR设计。最后，我们将带来新的结构预测技术和我们的 对实验室和临床同事正在进行的研究计划进行的蛋白质结构的统计分析 在福克斯·蔡斯癌症中心和坦普尔大学医学院。