image: Specializing in chemical and biological engineering, Bilodeau received her bachelor’s and master’s degrees from Northwestern University and her Ph.D. from Rensselaer Polytechnic Institute.
Credit: UVA School of Engineering
Camille Bilodeau, an assistant professor of chemical engineering at the University of Virginia School of Engineering and Applied Science, has earned a $600,000 CAREER Award from the National Science Foundation.
Bilodeau’s project explores the ways in which peptide molecules, which can bind to natural materials at the molecular level and influence biological processes, can be strategically tethered to surfaces and “tuned” for specific functions.
By leveraging molecular simulations and artificial intelligence, Bilodeau and her team are streamlining the complex process of designing peptide-covered surfaces for targeted applications, including development of new medicines, new technologies to desalinate water, and new approaches in semiconductor manufacturing.
Designing Peptides for Desired Properties
It might sound straightforward, but investigating how a peptide molecule, made of two or more amino acids, acts when attached to a particular surface is more complicated than it seems.
“There are 20 naturally occurring amino acids. If we are interested in designing a three-amino acid peptide, we have 20 to the third power, or 8000 options to choose from,” Bilodeau said.
“If, instead, we are interested in designing a 10-amino acid peptide, we have over a trillion options to choose from.”
Engineers regularly use molecular dynamics models and computer simulation to determine how they might design molecules and materials to produce desired properties that they can then apply to devices and other products.
“Molecular dynamics is literally taking all of the atoms in a molecule, calculating the forces and the accelerations on all of them, and stepping forward in time,” she explained. “Each step slightly shifts their positions, forming an ‘atomic movie’ that ultimately reveals how these atoms interact to create desired material properties.”
These applications could include designing a material for tissue grafting or organ tissue repair that has properties allowing it to toggle between sticky and non-sticky with a change in temperature, or a material that acts like a filter to remove toxins or impurities.
Exploring every possible peptide-surface interaction using molecular dynamics is impractical. The actual “tuning” of the patterns within a surface requires understanding how forces act upon the material and working those out for each combination, one by one. The process is computationally demanding, presenting an opportunity to leverage artificial intelligence.
“I'm going to attach this peptide to this surface to perform a specific function,” Bilodeau explained, while sketching an illustration on a dry-erase board. “But the challenge is that determining the right molecule takes a fair bit of time to do because we’re dealing with a lot of atoms. That’s where our deep-learning architectures come in.”
Early Research Group Milestones
“You can imagine that if you want to design a specific surface with a specific pattern, sifting through a thousand peptides searching for the one that's going to give you the properties that you're looking for would be enormously time-intensive,” Bilodeau said. “AI changes that.”
In her research group’s first paper, Bilodeau and colleagues introduced their integral deep learning model named PepMNet. In October, she spoke about their model at the American Institute of Chemical Engineers’ annual meeting, alongside peers using other AI solutions to speed up the prediction of materials properties.
That’s why the use of AI has become essential to Bilodeau’s work and that of her contemporaries.
“When a solution is needed quickly — say for a biohazardous spill that requires remediation, or for the next new drug that can deliver a targeted therapy during an epidemic — with new AI-driven insights, we hope to identify molecular solutions almost instantly,” she said.
As part of her CAREER award, Bilodeau and her group are using PepMNet to develop the first rapid predictive tool for understanding surface-water interactions of tethered peptides. If successful, it may lead to technological improvements in fields such as health and clean energy.
Benefiting Students, Industry Partners
Naturally, the graduate students in Bilodeau’s lab group will benefit from the NSF grant work. But, as part of the grant, her undergraduate students will as well, building their molecular understanding and scientific research skillsets based on the case studies from the NSF-driven projects Bilodeau presents in class as well as insights from her industrial and academic collaborations.
Bilodeau is no stranger to collaboration. She earned her doctorate from Rensselaer Polytechnic Institute in 2020 and, while a graduate student at RPI, received the Lawrence Livermore Advanced Simulations and Computation Graduate Fellowship. Through this fellowship she carried out research jointly between RPI and Lawrence Livermore National Laboratory.
Bilodeau’s group already has its first industry partner, BioRad Laboratories. Bilodeau previously worked with BioRad while completing her Ph.D. The company hopes to learn from the similarities between tethered peptides and the chromatography processes it uses in its existing drug purification technology, potentially improving purification.