SI2-SSE: GenApp - A Transformative Generalized Application Cyberinfrastructure

科学计算和计算分析正在成为几乎所有科学领域的组成部分，无论是物理、化学和生物学等精确科学，还是社会科学。许多研究实验室的工作重点是创建用于数据生成、分析和解释的科学代码。然而，在一个典型的研究实验室中制定的公共资助和往往来之不易的科学规范往往在资助的生命周期或工作人员轮转后不久就变得不可持续。那些仅仅为了维护和更新现有软件而提供昂贵的计算机科学专业知识的项目，从实验室的主要目标中转移了稀缺的资源，并且经常在没有解决问题的情况下转换问题。只有少数研究人员获得足够的资金来维护和更新软件，这限制了新思想和新技术的传播。软件环境的多样性和不断变化的性质使问题复杂化。启用用户利用率在部署、访问和培训方面存在障碍。这些问题也对执行新代码中体现的新想法造成了障碍。GenApp项目的目标就是解决这些问题。首先，GenApp能够快速将科学代码传播给具有最少软件专业知识的研究人员。随着越来越多的研究人员使用这些规范，越来越多的人被赋予了这些规范，这有助于它们的可持续性。该项目的基本目标是将GenApp框架推进为一种变革性工具，以广泛造福科学软件开发人员社区。GenApp是一个通用的应用程序生成框架，用于快速部署科学代码，它可以生成科学网关和独立应用程序。GenApp的主要独特功能之一是对最终用户的专业技术要求最低，以及确保生成的应用程序可持续性的开放式设计。为了生成功能齐全的应用程序，GenApp按照简单定义文件的指导，编织片段库和用户定义模块，这些文件是从GenApp提供的统一的、合乎逻辑的、易于编码的通用接口定义文件创建的。这个通用定义文件和底层软件可以无限地重用，以在各种现有的和尚未定义的软件环境中生成应用程序。在GenApp成熟的过程中保持这种简单性是主要的发展策略之一。为了实现GenApp的目标，提出了四个重点目标。第一个是基础设施开发，其中包括对GenApp功能的一般增强。第二个是文档、培训、传播、外联和可持续性——这些都是生产对社区有用的软件产品的重要方面。第三个目标是简单的反馈，因为用户和开发者的反馈将有助于推动前两个目标。最后的目标包括两个结构生物学领域的科学应用，将采用和推动GenApp的发展。GenApp的主要实际用途是为结构生物学的核磁共振（NMR）和小角度散射（SAS）领域的实验学家和理论家提供高要求的新型计算和分析工具。GenApp框架将作为应用程序的平台，利用需要高效利用HPC资源的先进工具、用分子模拟建模SAS数据的工具，以及一个大型软件套件，用于NMR和SAS测量的组合分析，并与计算建模相结合。方便地使用这些强大的工具将使迄今为止不可能的一些基本生物学问题的研究成为可能。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Scientific computing and computational analysis are becoming integral aspects of virtually any field of science, be it exact sciences like Physics, Chemistry and Biology, or social sciences. Efforts of many research laboratories are focused on creation of scientific codes for data generation, analysis and interpretation. However, publicly funded and often hard won scientific codes developed in a typical research laboratory too frequently become unsustainable beyond the lifetime of funding or shortly after staff rotation. Projects that are funded to afford expensive computer science expertise simply to maintain and update existing software divert scarce resources from the lab's primary goals and often translates the problem without resolving it. Only a select number of researchers receive sufficient funding to maintain and update software, limiting the dissemination of new ideas and techniques. The diversity and continually changing nature of software environments compounds the issues. Enabling user utilization presents hurdles in deployment, access and training. These issues also create barriers to the implementation of new ideas embodied in new codes. The GenApp project's goals are to address these issues. To begin with, GenApp enables the rapid dissemination of scientific codes to researchers with minimal software expertise. As more researchers use these codes, more of them become vested in the codes, which helps their sustainability. The fundamental goal of this project is to advance the GenApp framework into a transformative tool to broadly benefit the scientific software developer community. GenApp is a generalized application generation framework intended for rapid deployment of scientific codes, which can generate both science gateways and stand-alone applications. Among the main unique features of GenApp are the minimal technical expertise requirement for the end user and an open-end design ensuring sustainability of generated applications. To produce fully functional applications, GenApp weaves libraries of fragments and user defined modules as directed by simple definition files, created from a uniform, logical, and simple-to-encode general interface definition file provided by GenApp. This general definition file and the underlying software can be reused indefinitely to produce applications in a variety of existing and yet-to-be defined software environments. Preserving such simplicity with GenApp's maturation is one of the main developmental strategies. To achieve the goal of GenApp four focus Aims have been proposed. The first is infrastructure development, which includes general enhancements to the capabilities of GenApp. The second is documentation, training, dissemination, outreach and sustainability - all important aspects to produce a software product that is useful to the community. The third is simply feedback, since user and developer feedback will help drive the first two Aims. The final Aim includes two structural biology domain science applications that will adopt and drive GenApp development. GenApp will see its primary practical utilization in making highly demanding novel computational and analysis tools accessible to experimentalists and theoreticians working in the nuclear magnetic resonance (NMR) and small-angle scattering (SAS) domains of structural biology. The GenApp framework will serve as a platform for applications utilizing advanced tools requiring efficient use of HPC resources, tools for modeling SAS data with molecular simulations, and a large software suite for a combined analysis of NMR and SAS measurements coupled to computational modeling. Easy access to these powerful tools will enable hitherto impossible studies of a number of fundamental biological problems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.