Server / HPC Cluster

我们开发各种生物信息学领域的算法和工具，如网络生物学、遗传流行病学、系统医学等。这些帮助我们从分子数据和传感器数据中获得新的见解，并产生新的假设，可以用来探索各种疾病的新诊断和治疗方法。由于涉及的数据量很大，我们需要数百TB的存储空间。我们的许多工作也对计算硬件提出了非常高的要求，这是传统的网格计算体系结构无法满足的。例如，生物网络通常由数千个节点和数十万条边组成，需要特别大的主存储器才能进行充分处理。为了有效地调查这些数据和其他数据，同时需要大量的处理器核心。这种组合在传统的计算中心中没有表现出来，因为它们只提供许多处理器核心或大量的主内存。只有在高度并行的共享内存架构中，才能对分子和传感器数据进行更精确的基于人工智能的分析。由于不同项目对硬件需求的异构性，我们需要一个能够灵活满足这些需求的集群，并且根据项目的不同，可以提供足够的内存、处理器核、主存和GPU。

We develop algorithms and tools in various bioinformatics areas, such as network biology, genetic epidemiology, systems medicine, and others. These help us to gain new insights from molecular data and sensor data and generate new hypotheses that can be used to explore new diagnosis and treatment approaches for various diseases. Due to the amount of data involved, we need several hundred terabytes of of storage space. Much of our work also places very high demands on the computing hardware, which we cannot meet with conventional grid-computing architectures. For example, biological networks often comprise thousands of nodes and hundreds of thousands of edges and require a particularly large main memory to be processed adequately. In order to investigate these and other data effectively, a high number of processor cores is required at the same time. This combination is not represented in conventional computing centers, since they offer only either many processor cores or a large amount of main memory. More precise artificial intelligence- based analyses of molecular and sensor data can only be performed in highly parallel shared-memory architectures. Due to the heterogeneous hardware requirements in different projects, we need a cluster that can satisfy these needs flexibly and, depending on the project, can provide enough memory, processor cores, main memory, and GPUs, if necessary.