Data Management for Molecule Simulation : A Throughput-Oriented Approach

项目摘要 我们将设计和开发一个分子模拟数据管理系统，我们称之为P2 DMS，用于分析大规模的分子模拟数据。 分子动力学（MD）模拟数据。该系统的显着特点包括：（1）基于推送的本地 以批处理方式处理数据并同时处理多个查询的查询引擎设计; (2)使用现代众核硬件（如GPU）优化的MD分析工具;以及（3）有效的管理和 通过广域网访问分布式数据，这对于大规模MD模拟是非常常见的。这将 通过在最先进的分布式大数据管理系统之上构建数据分析层来实现。出去- 该项目的实施，不仅提高了MD数据处理的效率，而且使新的知识发现成为可能， 目前被认为是困难或不可行的。特别是，我们将把P2 DMS计划整合到现有的MD中， 模拟包，并验证新的设计与重要的现实世界的生物和MD方法学的问题， lems。特别是，我们将（1）模拟SARS-CoV-2刺突蛋白如何相互作用的结构-功能关系 与人血管紧张素转换酶2（ACE 2）受体作用;和（2）增强最近的 开发了MD模拟主动控制的参数优化软件。

Project Summary We will design and develop a molecular simulation data management system, we call P2DMS, to analyze large dis- tributed molecular dynamics (MD) simulation data. The salient features of the system include: (1) a push-based local query engine design that handles data in a batch processing manner and processes many queries at the same time; (2) optimized MD analytics tools using modern many-core hardware such as GPUs; and (3) efficient management and access to distributed data over wide area networks, which is quite common for large scale MD simulations. This will be done by building a data analysis layer on top of state-of-the-art distributed big data management systems. The out- come of this project will not only improve the efficiency of MD data processing, but also enable new knowledge discov- ery that is currently regarded difficult or infeasible. In particular, we will integrate the P2DMS program into existing MD simulation packages, and validate the new design with important real-world biological and MD methodological prob- lems. In particular, we will (1) model the structure-function relationships of how the spike protein of SARS-CoV-2 inter- acts with the human angiotensin converting enzyme 2 (ACE2) receptor; and (2) enhance the performance of a recently developed parameter optimization software for active control of MD simulations.