TLDR: Nabla Bio has announced a significant breakthrough in therapeutic antibody discovery with its AI system, JAM. This generative protein design platform enables the fully computational creation of antibodies with therapeutic-grade properties, addressing challenges in traditional and existing AI-driven antibody design methods. JAM has demonstrated success in designing antibodies for both soluble and hard-to-drug membrane proteins, including GPCRs, and has shown sub-nanomolar neutralization potency against SARS-CoV-2. The system significantly accelerates the design-to-characterization process to under six weeks, promising increased efficiency and access to previously intractable targets in drug development.
Nabla Bio, a pioneering company in the field of biotechnology, has unveiled its groundbreaking artificial intelligence system, JAM, marking a pivotal advancement in de novo antibody design. This innovative generative protein design platform is set to revolutionize therapeutic antibody discovery by enabling the fully computational creation of antibodies with high specificity, developability, and function, without the need for extensive experimental optimization.
Traditional methods and even earlier AI approaches in antibody design have faced significant limitations, including low hit rates and suboptimal affinities, often failing to produce high-quality lead antibodies robustly across various protein targets. Nabla Bio’s JAM system directly addresses these challenges, offering a robust solution that promises to expand the scope and efficiency of therapeutic antibody development.
According to Surge Biswas, Co-founder and CEO of Nabla Bio, “We’re incredibly excited to share new results from Nabla Bio where we show we can design antibodies de novo for use in therapeutic discovery.” The company has demonstrated JAM’s capabilities across a diverse range of targets, including soluble proteins and notoriously challenging multipass membrane proteins like GPCRs (G protein-coupled receptors), Claudin-4, and CXCR7. These membrane proteins are critical therapeutic targets for various diseases, including solid cancer tumors, and have historically been difficult to drug.
JAM-designed antibodies have shown impressive performance, achieving double-digit nanomolar affinities and strong early-stage developability profiles. Notably, against SARS-CoV-2, JAM-designed antibodies achieved sub-nanomolar pseudovirus neutralization potency, meeting established clinical benchmarks. The system’s ability to iteratively introspect on its outputs further enhances binding success rates and affinities, showcasing the potential for test-time compute scaling in physical protein design.
Also Read:
- Latent-X: A Breakthrough Generative AI Model Revolutionizes Protein Binder Design with No-Code Accessibility
- Absci Accelerates Drug Discovery with Major AI Investments
One of the most significant advantages of the JAM system is its accelerated timeline. The entire process, from computational design to recombinant characterization, requires less than six weeks. This dramatic reduction in development time, coupled with the ability to pursue multiple targets in parallel with minimal additional experimental overhead, positions JAM to unlock previously intractable targets and significantly increase the efficiency of therapeutic antibody development. This breakthrough aligns with the broader recognition of AI’s transformative power in predicting complex protein structures, a field recently acknowledged by the Nobel Prize in chemistry.
