Cybershuttle project to create seamless system to share data, tools
The National Center for Supercomputing Applications and the University of Illinois Urbana-Champaign are part of a five-year, $5 million collaborative award by the National Science Foundation to aid in the continued development of the Cybershuttle project, creating a seamless, secure and highly usable scientific research environment that integrates all of a scientist’s research tools and data, which may be on the scientist’s laptop, a computing cloud or a university supercomputer.
The Cybershuttle team integrates biophysicists, neuroscientists, engineers and computer scientists into a single team pursuing the project goals with a grounding in cutting-edge research problems such as understanding how spike proteins in viruses work, how the brain functions during sleep and how artificial intelligence techniques can be applied to modeling and design in engineering. The team also includes researchers from Indiana University, the University of California San Diego, the San Diego Supercomputer Center and the Allen Institute.
J.W. Hastings Endowed Chair and Professor of Biochemistry at the School of Molecular and Cellular Biology Emad Tajkhorshid is the lead principal investigator for UIUC. He is also the director of the National Institutes of Health Macromolecular Modeling and Bioinformatics Center, whose team will provide one of the main science applications of Cybershuttle.
“Combined with its experience in software development, the group seeks to take advantage of the computational infrastructure and workflows offered by Cybershuttle to streamline somewhat complex procedures involved in preparation, execution, and analysis of complex biomolecular and cellular systems.”
Emad Tajkhorshid, J.W. Hastings Endowed Chair and Professor of Biochemistry at the School of Molecular and Cellular Biology
Led by Tajkhorshid, the center’s main mission has been developing computational methodologies and implementing them in easy-to-use software for biomedical researchers. In addition, Tajkhorshid’s team has been involved in the modeling and simulation of complex systems such as membrane proteins and viral capsids and envelopes.
Cybershuttle will provide a centralized framework, helping users organize and automate the complex modeling and simulation workflows required in biophysics’ research, like the study of the spike protein of SARS-CoV-2 virus, for example. In addition, the framework will facilitate work reproducibility and data provenance, remote visualization for individual and collaborative work and streamline access to computational HPC resources.
But that’s not the only aspect of Cybershuttle impacted by UIUC researchers.
Like many computational fields, physics-based modeling is being revolutionized by the use of artificial intelligence (AI). A properly trained deep learning model can almost instantly produce results rivaling classical modeling methods without high-performance computing (HPC) resources or modeling software.
Researchers at NCSA and UIUC’s Grainger College of Engineering have recently developed and used such data-driven and physics-informed surrogate deep learning models to accelerate modeling and design in topological optimization, highly nonlinear material responses, turbulence, sensitivity and design, material processing and advanced manufacturing.
“Many parts of our AI confluence modeling workflow are currently executed manually, such as moving the data between low- and high-end computers,” said co-PI Seid Koric, associate technical director at NCSA and a research associate professor in Mechanical Science and Engineering. “Data preprocessing is additionally hampered by the required expertise in HPC and its batch and scripting environments.”
“Cybershuttle will be an ideal framework for the AI confluence modeling workflow that will enable the smooth use of diverse computational resources from laptops to HPC and back to laptops. Cybershuttle will facilitate the broader adoption of surrogate data-driven or physics-informed models in complex and computationally expensive physics-based modeling challenges in science and engineering, enabling high-quality modeling, optimization and design instantly accessible to scientists and engineers on low-end computing platforms such as laptops.”
Seid Koric, associate technical director at NCSA and a research associate professor in Mechanical Science and Engineering
Co-PI Jim Basney, a principal research scientist in NCSA’s Cybersecurity group, is the cybersecurity lead for Cybershuttle. The project will extend cybersecurity best practices used by middleware systems, such as science gateways, to the agent-enabled resources at the edge of the Cybershuttle hybrid architecture. Building on the NSF-funded Custos project, Cybershuttle will bundle federated authentication, authorization, fine-grained access control, data confidentiality, integrity and data sharing capabilities.
“Cybershuttle will leverage emerging fine-grained authorization mechanisms, such as SciTokens, for communicating with resources and providing auditable message trails to deliver a secure environment for trustworthy science,” Basney said.