Results
The A6T7 Chimera showed high predicted affinity for PFAS ligands, and six candidate enzymes were successfully expressed and sent off-site for further testing on PFAS substrates. We made an expression classifier (or regressor).
The A6T7 Chimera showed high predicted affinity for PFAS ligands, and six candidate enzymes were successfully expressed and sent off-site for further testing on PFAS substrates. We made an expression classifier (or regressor).
We harnessed protein large language models (pLLM) to generate novel PFAS degrading enzymes, and then computationally validated our enzyme’s activity with PFAS. Through many computational design, testing, and optimization runs, the team successfully created the A6T7 Chimera PFAS-degrading enzyme and 3 novel AI-designed dehalogenases. See Part:BBa_K5035003 and our Model page for more information.
We designed an expression classifier applicable to projects beyond our own, enhancing protein expression prediction for our, as well as future research. See our Software page for more information.
We were able to express our generated enzymes in the wet lab, later sending those enzymes to a separate lab to test them on a PFAS substrate. We are still waiting on results from our collaborator, but hope to be able to provide updates at the Jamboree. The success of our expression experiments allows us to make more informed computational predictions so that we can tailor our process to optimally create novel PFAS-degrading enzymes in the future.
By identifying different stakeholder groups including PFAS scientific experts, participants in the PFAS remediation industry, regulators and water treatment facilities, and community stakeholders, we could locate and engage with many experts. These experts influenced the direction of our project, resulting in pivots, enhancements, and validations at various stages.