Immunomodulators, small compounds with the potential to enhance vaccines and fortifying cancer immunotherapies, pose a challenge due to the vast number of drug-like molecules—estimated at 10^60, surpassing the stars in the visible universe.

In a groundbreaking development in vaccine design, machine learning facilitated the discovery of novel immune pathway-boosting compounds, singling out one small molecule that could surpass existing immunomodulators. The findings, featured in the journal Chemical Science, mark a significant advancement.

Immunomodulators“We employed artificial intelligence techniques to navigate an extensive chemical landscape, identifying molecules with unprecedented efficacy that traditional approaches might not have considered. We’re eager to share the methodology behind this,” stated Prof. Aaron Esser-Kahn, leading the experiments.

Prof. Andrew Ferguson, heading the machine learning efforts, noted, “While machine learning is prevalent in drug design, its application in immunomodulator discovery appears novel, showcasing the transferability of tools across different domains.”

Immunomodulators operate by altering the signaling of innate immune pathways. The team, led by Esser-Kahn, conducted a high-throughput screen exploring 40,000 molecule combinations, identifying top candidates. When these molecules were combined with adjuvants in vaccines, they demonstrated enhanced antibody response and reduced inflammation.

To expand the pool of candidates, the team integrated these findings with a library of nearly 140,000 commercially available small molecules, employing an iterative computational and experimental approach. Graduate student Yifeng (Oliver) Tang utilized active learning, a machine learning technique, to efficiently navigate molecular screening by learning from collected data and identifying high-performing molecules for experimental testing.

After four cycles, sampling only about 2% of the library, the team discovered small molecules with exceptional performance—increasing NF-κB activity by 110%, elevating IRF activity by 83%, and suppressing NF-κB activity by 128%. One molecule even induced a three-fold enhancement of IFN-β production when combined with a STING agonist, outperforming existing molecules by 20%.

The team also uncovered “generalist” molecules capable of modifying pathways when co-delivered with agonists, suggesting broader applications in vaccines. Understanding the common chemical features of these molecules enables a targeted approach to finding or engineering new compounds.

Continuing this process, the team aims to explore more molecules and encourages collaboration for enhanced dataset sharing. The focus is on screening molecules for specific immune activities and identifying combinations that offer better control over immune responses.

“Our goal is to discover molecules that can effectively treat diseases,” Esser-Kahn affirmed. The Pritzker School of Molecular Engineering at The University of Chicago utilized machine learning to guide high-throughput experimental screening in this innovative pursuit.

Other posts

  • AI-Powered Human Resources
  • Emotion Recognition with Machine Learning
  • Google Introduces The Open Source Framework Firebase Genkit
  • AI Innovations Presented at Google I/O
  • Researching Genetic Algorithms
  • Yelp Introduces an Innovative AI Assistant to Facilitate Business Connections
  • The Intersection of AI and Machine Learning in Financial Services
  • Anomaly Detection Using Machine Learning Algorithms
  • Meta Debuts The Latest AI Chip In A Competitive Technology Sprint
  • An Innovative Machine Learning Approach for Analyzing Material Surfaces