找回密码
 新注册用户
搜索
查看: 3051|回复: 2

OpenPandemics进入beta测试

[复制链接]
发表于 2020-4-22 02:50:20 | 显示全部楼层 |阅读模式
北京时间21日早上开启beta,目前为手动分发测试包,主要测试AutoDock平台、结果验证以及任务包大小,过段时间会转为自动分发
quick update: we just started an initial Beta test. For now, we’re using manually created work units, which are actually targeting a protein in the virus. The aim of this initial beta run is to test the science application (AutoDock), results validation and work unit sizing. The initial validation will be with redundancy (where we send out two copies of each work unit), which will help us confirm the validator works against all CPU types within the same operating system. However, at launch, this project should run similarly to other previous AutoDock-based World Community Grid projects - i.e. besides random comparison checks, work units will run with no comparison/redundancy.

In the meantime, we are testing the automation of work unit generation on our Alpha environment and once those tests are complete, we’ll proceed with automatic work unit generation on beta.
另外项目专家分享了为何使用旧版的AutoDock 4而不是更高效率的AutoDock Vina,有兴趣的各位自行阅读~
First of all, thanks for the interest in our project! This post is to elaborate on why we opted for AutoDock4 instead of AutoDock Vina.

Both Vina and AutoDock4 are developed in our lab under the AutoDock Suite project. Each has their own advantages, and that is why we keep working with both programs. Generally, docking involves evaluating thousands to millions of conformations and orientations (i.e. poses) of the molecule being docked, before the best result is found. While evaluating a single pose takes approximately the same time in both AutoDock4 and Vina (a few microseconds), Vina is faster because it needs to evaluate fewer poses before reaching the best result. The reasons for this are an efficient search algorithm and a simpler scoring function to describe the interactions.

What Vina gains in performance is lost in flexibility. AutoDock4 allows expert users to tweak the scoring function and define custom potentials. This allows the implementation of very complex protocols and functionalities that would not be possible otherwise. One of these protocols is the Reactive Docking, which is used to model molecules that are able to react with the protein and form a covalent bond. Such molecules are often called suicide inhibitors (a well-known example is penicillin). AutoDock4 can be used to simulate the formation of a chemical bond between a reactive molecule and a protein, thereby predicting if the reaction occurs. A neat feature is that we can implement this protocol without requiring any changes to the code of the program that runs on the WCG, which reduces considerably the effort to use this method without extensive coding efforts on either our end or the IBM side. The Reactive Docking is one of the aspects that make this project unique, because it will allow us to identify suicide inhibitor candidates to be tested by our collaborators.

Besides, for this project we are tuning the search parameters to reduce the number of poses that AutoDock4 evaluates. Our tests show that for some molecules this will bring the runtime of AutoDock4 closer to that of Vina.


回复

使用道具 举报

发表于 2020-4-22 10:10:25 | 显示全部楼层
OpenPandemics 就是研究 COVID-19 的吧?
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 新注册用户

本版积分规则

论坛官方淘宝店开业啦~
欢迎大家多多支持基金会~

Archiver|手机版|小黑屋|中国分布式计算总站 ( 沪ICP备05042587号 )

GMT+8, 2024-3-28 18:54

Powered by Discuz! X3.5

© 2001-2024 Discuz! Team.

快速回复 返回顶部 返回列表