On September 16, 2026 Nona Biosciences ("Nona"), a global biotechnology company advancing biologics discovery and development through innovativetechnology platforms, reported the development of HCAbLM, the world’s first language model specifically trained on fully human heavy-chain-only antibodies (HCAbs). HCAbLM was built on a large-scale repertoire comprising 31.8 million fully human HCAb sequences from 73 independently immunized HCAb transgenic mice. The study, "A foundation model learns the sequence and functional grammar of fully human heavy-chain-only antibodies," was published in bioRxiv.
Schedule your 30 min Free 1stOncology Demo!
Discover why more than 1,500 members use 1stOncology™ to excel in:
Early/Late Stage Pipeline Development - Target Scouting - Clinical Biomarkers - Indication Selection & Expansion - BD&L Contacts - Conference Reports - Combinatorial Drug Settings - Companion Diagnostics - Drug Repositioning - First-in-class Analysis - Competitive Analysis - Deals & Licensing
Schedule Your 30 min Free Demo!
From Binder to Drug Candidate: A Leap in AI Antibody Development
The main bottleneck in the current AI antibody field is no longer "whether an antibody that binds the target can be designed," but rather "whether the designed antibody can be manufactured at scale, whether it will aggregate, and whether it is stable"—that is, developability. HCAbLM was created precisely to address this issue. The model has learned the "sequence gramma" unique to fully human HCAbs—a set of underlying rules that determine whether an antibody can fold, exist stably, and ultimately become a drug. Conventional general-purpose models cannot capture this set of rules because their training data contain almost no HCAb sequences. HCAbLM fully demonstrates that a dedicated model trained on unique antibody sequence data can better translate "sequence space cognition" into "developability prediction."
A Dedicated HCAb Model Outperforms General-Purpose Protein and Antibody Models
The representations learned by HCAbLM demonstrate transferability to experimentally measured antibody properties, including size exclusion chromatography (SEC) purity, hydrophobic interaction chromatography (HIC) behavior, and thermal stability, highlighting HCAbLM’s potential in the evaluation and optimization of fully human HCAbs, going beyond mere sequence analysis.
In public benchmarks for cross-project developability prediction, HCAbLM, with only 366 million parameters, surpassed general-purpose protein large models including Meta’s ESM-6B with 6 billion parameters, as well as previously leading antibody language models IgLM and AbLang, on two core metrics: SEC purity and HIC behavior.
The development of HCAbLM marks an important advancement in Nona’s application of artificial intelligence (AI) to antibody discovery and development. By combining large-scale antibody repertoires with AI-driven modeling, Nona is building a foundation for more efficient and data-driven antibody design and developability assessment.
"The development of HCAbLM represents an important step in our efforts to apply AI to specialized antibody formats and unlock the value of large-scale antibody repertoires," said Dr. Di Hong, Chief Executive Officer of Nona Biosciences. "By learning the unique sequence characteristics of fully human HCAbs and demonstrating transferability to experimentally measured molecular properties, HCAbLM provides a new foundation for AI-enabled antibody discovery and development. We look forward to further integrating HCAbLM with Nona’s technology platforms, and leveraging these capabilities to accelerate antibody discovery and development while empowering our partners to unlock new opportunities for next-generation biotherapeutics."
(Press release, Nona Biosciences, SEP 16, 2026, View Source [SID1234670896])