论文部分内容阅读
化学计量学被广泛应用于光谱、色谱及质谱数据处理,现有化学计量学软件都为串行单机程序,这将导致程序开发成本高、部署升级困难、可控制性差等缺点,难以胜任分析数据的成倍增长对快速分析和有效管理带来巨大的挑战。为了解决这些缺点,本文提出了CloudChem——1种基于云计算的化学计量学软件服务,它采用软件即服务的模式,使用浏览器/服务器结构来提供专业的化学计量学软件服务。系统的服务器端分为工作流层、业务层、计算层、数据层和Web层。数据层综合利用关系型数据库和分布式文件系统2种方法的优势来存储和组织数据;计算层使用并行计算来提升数据处理速度和规模;Web层使用Open API实现平台的数据资源共享功能。实验表明,平台上的并行交叉验证框架在四核CPU上加速比可以达到3.763倍。所以,CloudChem可以较好克服传统化学计量学软件的缺点,基于该方法的软件服务平台可实现光谱、色谱、核磁、质谱等数据的有效、高速、一体化的存储、分析、挖掘,最大限度减小用户在使用化学计量软件的基础设施成本和软件成本。
Chemometrics is widely used in spectrum, chromatographic and mass spectrometry data processing. Existing chemometrics softwares are all serial single-machine programs, which will lead to high cost of program development, difficult deployment and upgrade, poor controllability, etc., which make it difficult to analyze the data The doubled growth presents a huge challenge to rapid analysis and effective management. To address these shortcomings, this paper presents CloudChem, a cloud computing-based chemometrics software service that takes a software-as-a-service model and uses a browser / server architecture to provide specialized chemometrics software services. The server side of the system is divided into workflow layer, business layer, computing layer, data layer and Web layer. The data layer comprehensively utilizes the advantages of the relational database and the distributed file system in two ways to store and organize the data; the computation layer uses the parallel computation to improve the data processing speed and scale; and the Web layer uses the Open API to realize the data resource sharing function of the platform. Experiments show that the parallel cross-validation framework on the platform can achieve 3.763 times faster than the quad-core CPU. Therefore, CloudChem can overcome the shortcomings of traditional chemometrics software. The software service platform based on this method can effectively store, analyze and mine spectrum, chromatogram, nuclear magnetic resonance and mass spectrometry data, and minimize Small users are using infrastructure software and stoichiometric software costs.