By Cherie Saffold
The Vanderbilt University Center for Structural Biology and Center for AI in Protein Dynamics recently hosted a symposium called “Bridging Atoms and Algorithms: A Symposium on AI and Structural Biology.” The March event brought together Vanderbilt faculty, trainees, and invited speakers to explore the intersection of artificial intelligence and structural biology, share research, and foster new scientific collaborations.
The symposium, which was co-sponsored by the SBGrid Consortium, opened with workshops designed to introduce attendees to emerging tools in artificial intelligence and cryo-electron tomography, a 3D imaging technique used to resolve molecular structures in situ.
The AI workshop began with an introduction from Hassane Mchaourab, Louise B. McGavock Professor, professor of molecular physiology and biophysics, and director of CAIPD. He outlined the center’s mission to leverage artificial intelligence to better understand disease mechanisms and advance human health. A central focus of CAIPD’s work is integrating AI with experimental data to improve protein dynamics modeling. One such effort, DEERFold, builds on AlphaFold, an AI program that predicts the structures of proteins, by incorporating experimental constraints to refine protein structure predictions. CAIPD is also collaborating with researchers to curate a large, gold-standard database of experimentally determined protein conformations.
Following this introduction, CAIPD researchers presented ways to use AI responsibly and effectively in research projects. Assistant Professor of Molecular Physiology and Biophysics Carlos Oliver described UNICORNE, a database of experimentally solved protein conformations that enables researchers to explore protein dynamics and structure prediction from primary sequences. Derek Claxton, also an assistant professor of MPB, discussed computational approaches for identifying disease-associated mutations, emphasizing the role of variant effect predictors—bioinformatics tools used to estimate how genetic variants influence disease phenotypes. He noted the importance of understanding how these models are trained to interpret their predictions effectively. Dr. Benjamin Brown, assistant professor of pharmacology, highlighted the use of machine learning to predict drug-protein interactions, with applications in accelerating drug discovery.
In parallel, the cryo-ET workshop, sponsored by Thermo Fisher Scientific, featured presentations highlighting advances in imaging technologies and data acquisition.
The symposium also featured invited speakers from across the field. Rommie Amaro, a professor of molecular biology at the University of California San Diego, delivered the keynote address, “Multiscale Computational Microscopy for In Situ Molecular Dynamics.” She presented her work on computational microscopy, which integrates simulations across multiple time and length scales to characterize protein behavior in complex environments. Her group has applied these approaches to model SARS-CoV-2, revealing how spike protein glycans contribute to immune evasion. This work illustrated the growing synergy between structural biology and computational modeling in generating predictive insights into protein function.
Additional presentations highlighted the breadth of current research in the field. Danielle Grotjahn, associate professor of integrative structural and computational biology at Scripps Research, described new AI-based approaches for reconstructing complex membrane architectures from cryo-ET data. Her work examined how mitochondrial membranes remodel during stress and replication. Invited researchers such as Joey Davis, associate professor of biology at MIT, and Greg Bowman, the Louis Heyman University Professor at the University of Pennsylvania, further explored how data-driven models can capture protein dynamics and inform our understanding of disease-associated mutations.
Vanderbilt researchers also presented work spanning conformational ensembles, bacterial toxin release mechanisms, drug discovery, antibody epitopes, and membrane protein dynamics. Presentations from Brown, as well as fellow faculty members Stephanie Wankowicz, Shannon Kordus, Ben Spiller, and Qiangjun “QJ” Zhou, and Mchaourab lab postdoc Qingyu Tang highlighted the wide range of structural and computational biology research taking place across Vanderbilt and Vanderbilt Health.
Beyond the scientific sessions, the symposium provided opportunities for community building and collaboration. A reception sponsored by Thermo Fisher Scientific followed the keynote address and featured a science Jeopardy game hosted by Associate Professor of Medicine Ray Blind. The event concluded with a poster session showcasing research from trainees, faculty, and CSB core facilities.
The symposium underscored the increasing integration of structural biology and computational techniques. By combining high-resolution experimental techniques with advances in AI, researchers are gaining new insights into protein dynamics, improving drug discovery efforts and advancing our understanding of how genetic variation contributes to disease.
