中国科学软件网-首页
GAMS软件制药行业案例——CyBio调度器

CyBio调度器-用于高通量筛选调度软件

高通量筛选是一种广泛应用于药物研究特别是在药物发现领域的科学实验方法。因为大量有前景的新药难以用人工分析,筛选过程是使用机器人自动化的。自动筛选系统用来处理含有化合物的微孔板。这些自动筛选系统可以在给定的一组微板上执行一系列任务和实验(称为检测方案)并生成实验数据。


CyBio在2009年并入到Analytik Jena AG中,马克斯普朗克研究所马格德堡开发了利用GAMS的优化方法来增加机器人筛选系统的产量。GAMS驱动检测方法的优化极大地提高了高通量筛选系统的生产速度并改进了实验数据的质量。


问题

GAMS的分析优化是CyBio调度器的一个不可或缺的部分,在这之前,只有专家能够修改时间以提高吞吐量。这种方法只适用于相对较小的试验而且是一项需要数小时来专注工作的任务。随着检测协议的日益复杂,这项任务现在已经远远超出了人类所能处理的范围。

另一个常见问题是一个代数模型描述筛选系统以减少空闲时间为重点之前,筛选系统利用关键资源方面效率底下。由于沉积、衰变或温度漂移,化合物的闲置时间也会导致实验数据的系统误差。


设置

CyBio Scheduler的中心部分是用GAMS编写的代数模型.它描述了一种让所有部件空闲时间最小化的筛查系统从而保证关键资源的最有效利用。可以使用多个资源来完成不同的任务,因此,筛选系统可以同时使用其他闲置设备来处理大量的微板。短而直接的微板促进资源高效利用,从而提高了生产效率。该模型在协调资源访问时避免了冲突,并确保所得到的调度是未锁定的。


许多制约是系统固有的,比如有限的临时存储或不能同时使用的资源和必须进行访问的资源。某些制约因素是特定于分析的。通常,用户定义了潜伏期的目标时间,包括上下边界,或者特定事件之间的最大时间间隔。因此,例如,复合添加和测量之间的时间可能是有限的。分析定义和这些制约创建了一个分离不等式系统。

由于严格的时间要求,微板在每个周期都遵循相同的进程。快速而均匀的微板处理与CyBio调度器减少了沉积、衰减或温度漂移所带来的系统误差,这是很难量化的。因此,增加的吞吐量不仅减少了每个实验的投资,而且还改善了数据质量。


用户友好性

用户可以从系统布局决策中解脱出来,并可以专注于实验。随着GAMS模型在后台运行,CyBio Scheduler着重于提供一个简单且方便的用户体验。它隐藏了将分析协议映射到当前系统设计的复杂性并且为目标找到全局最优解,以最小化循环时间。

用户可以从系统布局决策中解脱出来,并且可以专注于这个实验。CyBio Scheduler可以在它们需要的地方自动插入微板传输,解决资源分配冲突的问题,允许潜伏并且可以轻松的指定。根据独立任务的数量、涉及的组件和约束,所生成的模型可能相当复杂。最优解通常计算得足够快,才可以让用户验证是否可以放松某些约束,以便得出更好的结果。

Robotic screening systems are used to handle microplates containing chemical compounds. These robotic screening systems perform a sequence of tasks and experiments on a given set of microplates – called the assay protocol – and generate experimentation data.


CyBio, merged into Analytik Jena AG in 2009, and the Max Planck Institute Magdeburg developed optimization methods involving GAMS to increase the throughput of robotic screening systems. The GAMS driven assay optimization has significantly boosted the production rate of high throughput screening systems and improved the quality of the experimentation data.


The Problem

Before assay optimization involving GAMS was an integral part of CyBio Scheduler only experts were able to modify the timing to improve the throughput. This was feasible only for relatively small assays and was a task that involved hours of focused work. With growing complexity of assay protocols, this task is nowadays far beyond what human labor can handle.

Another common issue before an algebraic model described the screening systems with a focus on reducing idle time, were inefficiencies in the utilization of critical resources. Idle time of the compounds also leads to systematic errors in the experimentation data due to sedimentation, decay, or temperature drift.


The Setup

A central part of the CyBio Scheduler is an algebraic model written in GAMS. It describes the screening systems in a way that allows the minimization of idle time for any component ensuring the most efficient utilization ratio for critical resources. Several resources may be used for different tasks, so it is possible for the screening system to simultaneously process a number of microplates using else idle devices. Short and direct microplate transfers facilitate an efficient resource usage and thereby a high production rate. The model avoids conflicts when coordinating resource access and ensures that the resulting schedule is deadlock free.


A number of constraints are inherent to the system, such as limited temporary storage or resources which cannot be used simultaneously and to which access must be coordinated. Some constraints are assay specific. Typically the user defines the target time for incubation periods including an upper and lower bound, or the maximum time between specific events. So, for example, the time between compound addition and measurement may be limited. Assay definition and these constraints build a system of disjunctive inequalities.


Due to the strict timing, micro plates follow an identical itinerary for each cycle. Fast and uniform microplate processing with the CyBio Scheduler reduces systematic errors introduced by sedimentation, decay, or temperature drift, which are difficult to quantify. An increased throughput therefore not only reduces the investment per experiment but also improves the data quality.


User Friendliness

With the GAMS model running in the background, the CyBio Scheduler focuses on providing a simple and convenient user experience. It manages to hide the complexity of mapping an assay protocol to the current system design and finding the global optimal solution for the objective to minimize the cycle time.

The user is relieved from system layout decisions and can focus on the experiment. The CyBio Scheduler automatically inserts microplate transports where they are required, resolves conflicts in resource allocation and allows for incubations to be effortlessly specified. Depending on the number of independent tasks, involved components and constraints the resulting model may become considerably complex. However, the optimal solution is typically calculated fast enough to allow the user to verify if relaxing some constraints may lead to a better result.