Chemistry and Computer Science Collaboration Wins NSF Grant
Content
UMass Amherst researchers have received a $500,000 National Science Foundation (NSF) grant for a project aimed at improving how scientists from various areas identify unknown molecules through spectroscopic analysis.
The project, entitled “ACED: Revolutionizing Instrumental Analysis Using Foundation Models,” is being led by principal investigator Zhou Lin from the College of Natural Sciences’s Department of Chemistry and co-investigator Hui Guan from the Manning College of Information and Computer Sciences. By combining chemical knowledge with state-of-the-art foundation models in artificial intelligence (AI)—especially large language models (LLMs)—Lin and Guan aim to create a universal toolkit that can automatically and accurately translate complex spectral signals into detailed molecular structures without trial and error. This would dramatically accelerate chemical analysis across multiple fields.
“This project is not just about scientific advancement; it's about making a real difference in our world,” argues Lin. “By addressing a significant bottleneck in spectroscopic analysis—namely the slow and specialized trial-and-error conversion of raw spectroscopic signals into atomistic molecular structures—we're paving the way for faster and more accurate chemical discovery. The resulting open-source toolkit will make advanced spectroscopic analysis more accessible, benefiting academia, industry, and public agencies in Massachusetts and beyond.”
Currently, when a scientist wants to identify mysterious substances in an unknown sample of solid, liquid, or gas, they often use a spectroscopic machine that shines light to measure it and creates a numerical sequence of signals, known as a special molecular "fingerprint." This requires considerable time and skill to manually convert a numerical fingerprint into the molecular structure of the mysterious substance.
“Our project utilizes powerful AI—particularly the technology behind Google Translate—to instantly ‘translate’ a numerical fingerprint into an atomic-level picture of the unknown molecule,” explains Lin. “Our toolkit will significantly accelerate the solution of chemical mysteries in various areas.”
Lin and Guan see the potential applications of this research as being vast and varied. The proposed AI toolkit will transform any field that relies on fast spectroscopic identification of molecular structures. Beyond chemistry-related areas, it has the potential to revolutionize: healthcare (non-invasive diagnoses); pharmaceutics (drug discoveries and quality control); environmental science (pollutant detection); food safety (contaminant and spoilage measurement); national security (hazardous chemical and explosive detection); astronomy (molecular identification in space); and education (accessible spectroscopic analysis in teaching labs).
“This NSF award is not just another research project,” expressed Lin. “It marks the beginning of a new research direction, where AI accelerates scientific discovery in a distinct area.”
Lin and Guan’s first paper related to this grant is now under revision.
Learn more about this NSF grant.