Research Starter Grant: New Physics-Based Approaches to Predictive Protein Modeling, with Applications to Cellular Regulation

该项目的目标是开发基于全原子物理模型的蛋白质结构预测的新计算方法，具体目标是提供新的计算工具来研究细胞调控的结构机制。尽管有许多全原子物理化学模型被用来阐明特定蛋白质的功能，但这种模型目前在预测建模中发挥的作用非常有限。从算法开发的角度来看，重点是实施新的采样方法，以补充分子动力学，从合适的初始猜测(例如，同源模型)有效地探索蛋白质的能量格局，以及验证和改进用于蛋白质结构预测的能量模型。具体的应用包括同源建模，预测磷酸化引起的结构/动力学变化，以及预测多结构域蛋白质的堆积排列。

The objective of this project is to develop new computational methods for protein structure prediction that are based on all-atom physics based models, with the specific goal of providing new computational tools to investigate the structural mechanisms of cellular regulation. All though there are numerous examples of all-atom physical chemical models that are being used to elucidate the function of specific proteins, such models currently play a very limited role in predictive modeling. From the standpoint of algorithmic development, the central emphasis is on the implementation of new sampling methods, which complement molecular dynamics, for efficiently exploring the energetic landscape of a protein starting from a suitable initial guess (for example, homology model), and on the validation and refinement of energetic models for protein structure prediction. Specific applications include homology modeling of kinases, the prediction of structural/dynamical changes due to phosphorylation, and prediction of the packing arrangements of multi-domain proteins.