ARTIFICIAL INTELLIGENCE-ASSISTED INTEGRATIVE BIOMOLECULAR NMR PLATFORM

Understanding functions through structural conformations, dynamics, and allostery becomes more important because only one snapshot of the energy landscape is often insufficient to understand biological phenomena. Nuclear magnetic resonance (NMR) spectroscopy is the most suitable method for studying functions through structural conformations, dynamics, and allostery, which are missing from a simple structural snapshot obtained via other methods.

However, complex data interpretation and the lack of automation have been barriers to the adoption of the NMR method. This lack of automation often stems from poor and complex data quality, which hampers the successful automation of individual analysis steps. Further, the larger biological community has been hesitant to adopt powerful NMR methods due to a steep learning curve, which partly stems from the lack of a user-friendly end-to-end software platform.

With this award, the POKY team provides an improved software environment for biomolecular NMR studies by developing and employing an integrated artificial intelligence (AI) platform around the POKY software suite. With an AI-assisted NMR software platform, broader impacts will be achieved by facilitating investigations into biomolecular structure, function, and the mechanisms of molecular machines, primarily composed of proteins.

This understanding is crucial for addressing challenges related to environmental preservation, feeding growing populations, and promoting health and welfare by combating diseases. Underrepresented scientists from regional and international groups will learn and utilize the platform through social networks and virtual technologies. Various outreach events are planned to demonstrate how computational technologies can be leveraged to rapidly understand biological phenomena.

NMR spectroscopy, unlike other major methods like X-ray crystallography and cryoelectron microscopy (cryo-EM) used for determining protein structures at atomic or near-atomic resolution, enables the study of protein structures and dynamics at the atomic level in various physiologically relevant environments. Both solution and solid-state NMR methods provide valuable and complementary information.

The objectives of this project are to develop an integrated and automated analysis POKY platform, employ advanced computational techniques, and disseminate the software tools to the public for these NMR methods for both novice and expert users. The first goal is to develop an automated and integrated analytical platform for biomolecular NMR. The improved analytical platform will utilize AI/ML/CV (artificial intelligence/machine learning/computer vision)-based techniques for chemical shift assignments, AI/ML/CV-assisted homomer structure determination, and provide for a user-friendly interface (UI) and user experience (UX) for NMR-based structure and dynamics studies.

A second goal is to develop and employ advanced computer techniques in biomolecular NMR. Advanced computer science techniques are developed and employed by a new multidimensional NMR data format, including CV-based spectral quality enhancement and implementation of new workflows using CV/ML techniques. A third goal is to effectively distribute deliverables to push biomedical sciences forward.

Deliverables are disseminated efficiently by developing and providing web servers eliminating the need for large computational resources and complicated installation steps or active user engagement and by expanding plugin repositories and virtual environments. Scientists will reap significant benefits from the full-stack development and widespread adoption of the intuitive software suite POKY.

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.