【作者】 黄钰；

【导师】 骆清铭；

【作者基本信息】 华中科技大学 ， 生物医学工程， 2008， 博士

【摘要】 在生命科学领域内,新的高效的实验技术产生了海量的生物学数据,针对这些数据的分析也产生了大量的算法和软件工具。异构的数据源阻碍了不同实验室之间的数据交换,不同的算法软件之间也缺乏统一的信息交互标准,如此大量的数据和算法在相关的各个领域都需要有效的整合。基于此,生命科学的相关领域具有集成功能的软件包被不断开发出来。然而,第一,这些相关领域的软件包都没有提供通用一致的数据和算法接口。第二,这些相关领域的软件工具都没有提供协同工作机制。第三,这些软件工具无法在不同的操作系统间无缝的移植。此外,它们普遍缺乏友好的客户端图形操作界面GUI(graphical user interface)。本文提出构建一个基于Java环境的工作流管理平台BioW(Bio-Workflow)。该平台基于工作流技术,采取Java 2平台企业版(Java 2 platform)的体系结构。它不仅(1)给出通用一致的接口来集成算法工具和数据源;(2)其工作流机制使得被集成的软件更容易进行信息交互。BioW平台所提供的独立软件开发包software development kit(SDK)包含通用的数据接口Bio-data,用户可以利用该接口集成各种异构数据源,譬如神经科学领域的各种神经元序列等。该SDK还包括一致的算法接口Bio-node,用户可以利用该接口集成各种形式的算法工具,包含一系列开源的Java代码,开源的其他代码(C,C++等),以及可执行的二进制程序。其次,BioW提供多用户管理模块和协同工作机制。基于此,每个用户都拥有各自的空间存储自己的数据和分析工作流。此外,用户组的公共空间所存储的数据和工作流可以被同组的其他用户共享,这使得研究者可以无冲突的共享实验数据和分析过程,提高了复杂研究工作的效率。再次,该平台提供友好的客户端图形界面GUI,研究者可以在该界面上以拖拉节点的方式来定制,修改和执行工作流程。最后,BioW基于Java构建,因此独立于操作系统,整个工程一次编译后便可部署于不同的操作系统。更多还原

【Abstract】 In life science, new techniques and experimental methods produce explosive amounts of biological information. Accordingly, to analyze the experimental data, many algorithms and computer programs have been developed. Since these programs are developed by commercial companies or academic institutes independently, the lack of common standard makes the talks between them difficult. In one integrative analysis, when several of them are needed, the researchers are usually exhausted by the program switching and data transformation. Firstly, these software tools in corresponding field did not provide the standard data interface and algorithm interface. Secondly, all the tools and software mentioned above did not support collaborative analysis on the experimental data. Additionally, these tools and software are not operation system (OS) independent. Furthermore, the lack of a friendly interface, such as graphical user interface (GUI), will impede the efficiency of data analysis.In this paper, we developed an integrative workflow-based platform for the experimental data analysis, namely BioW (Bio-Workfow). BioW is built on the Java 2 Platform Enterprise Edition (J2EE), which provides (1) a general application development interface to integrate or bridge other programs and (2) a workflow mechanism to operate them and make them talk easily.The SDK provided by BioW defines a series of interface, such as Bio-data and Bio-node, with which the users can integrate any data sets, routines and viewers into the neural spike train data analysis without caring their input-output data formats, programming language and OS. The current existed algorithms (viewers) are classified into three categories: open source of algorithms programming with Java, open source of algorithms programming with other language (such as C, C++, etc.) and binary executable algorithms programs. The users can easily integrate all these kinds of algorithms into the platform through the interface Bio-node defined by SDK.The user management mechanism provided by BioW support collaborative mechanism, which enables multiple users share the same data and the same workflow without conflict. According to this mechanism, each user has a private user space where workflows and data are stored, and a public user space that be shared with other users belonging to the same group. With this sharing mechanism, the different users in the same group can not only share the same data, but also edit and execute the same workflow. Therefore many users can work together on the same experimental data analysis. This collaborative mechanism is necessary for the researchers to handle some complex neuroscience project.Moreover, as a Java-based software, BioW can be deployed in different OS without any modification, such as recompiling the whole project. Friendly GUI that provides the construction and management of workflows greatly facilitates the managing and processing the experimental data.更多还原