Open Mx: Multipurpose Software for Statistical Modeling

描述(由申请人提供)：摘要统计建模在广泛的科学调查中发挥着核心作用。对复杂特征和疾病的研究，如成瘾行为、精神病理学、心血管疾病、肥胖和癌症，现在面临着一系列需要改进软件的统计挑战。挑战包括：i)测量行为特征的困难；ii)可用技术--如磁共振成像、连续生理监测和微阵列--产生非常大量的数据，通常具有复杂的随时间变化的模式，以及iii)用于分析数据的统计模型更加复杂。目前的项目提议重写广受欢迎的Mx统计资料包，以应对这些挑战。模型的用户规格和用于测试它们的算法都将得到显著改进。该软件将：i)分成与R统计包互操作的模块；ii)作为开放源码发布，以便为今后的维护和开发提供一条稳定的途径；iii)与VDL并行工作软件集成。通过使用提供详细来源跟踪和派生结果注释的分析工作，使用VDL的网格/并行计算和数据管理将大大加快处理大型(高达数TB)数据集的速度。修改的模型和优化算法将增加软件的范围和性能。将在国家和国际讲习班上记录和传播该守则及其使用情况。对复杂特征和疾病的研究，如成瘾行为、精神病理学、心血管疾病、肥胖和癌症，现在面临着一系列需要改进软件的统计挑战。挑战包括：i)测量行为特征的困难；ii)可用技术--如磁共振成像、连续生理监测和微阵列--产生极大量的数据，通常具有复杂的随时间变化的模式；iii)用于分析数据的统计模型更加复杂。目前的项目建议开发使用大规模并行计算网格的软件，即“网络基础设施”，以应对这些挑战。

DESCRIPTION (provided by applicant): Summary Statistical modeling plays a central role in a wide range of scientific investigations. Studies of complex traits and disorders such as addictive behaviors, psychopathology, cardiovascular disease, obesity, and cancer are now faced with a set of statistical challenges that require improved software. The challenges include: i) the difficulty of measuring behavioral traits; ii) the availability of technologies - such as magnetic resonance imaging, continuous physiological monitoring and microarrays - which generate extremely large amounts of data often with complex time-dependent patterning, and iii) increased sophistication in the statistical models used to analyze the data. The current project proposes to rewrite the popular Mx statistical package in order to address these challenges. Both the user specification of models and the algorithms used to ?t them will be significantly improved. The software will be i) split into modules that interoperate with the R statistical package, ii) released as open source so as to provide a stable path for future maintenance and development and iii) integrated with the VDL parallel work software. Grid/parallel computing and data management using VDL will provide significant speedup for processing large (up to multi-terabyte) data sets, through the use of analytical work that provide detailed provenance tracking and annotation of derived results. Revised algorithms for model and optimization will increase both the scope of the software and its performance. Both the code and its use will be documented and disseminated at national and international workshops. Studies of complex traits and disorders such as addictive behaviors, psychopathology, cardiovascular disease, obesity, and cancer are now faced with a set of statistical challenges that require improved software. The challenges include: i) the difficulty of measuring behavioral traits; ii) the availability of technologies - such as magnetic resonance imaging, continuous physiological monitoring and microarrays - which generate extremely large amounts of data often with complex time-dependent patterning; and iii) increased sophistication in the statistical models used to analyze the data. The current project proposes to develop software that uses massively parallel computing grids, "cyberinfrastructure", in order to address these challenges.