SandboxAQ’s protein engineering platform combines AI, molecular simulation, and computational chemistry to accelerate antibody design—offering fast, accurate affinity predictions.
Antibody discovery remains a high-stakes, resource-intensive process. Despite advances in biologics, R&D teams and antibody program owners still face persistent bottlenecks—slow iteration cycles, incomplete data at key decision points, and predictive models that struggle with structurally disordered regions like complementarity-determining regions (CDRs).
Biopharma must manage the risks and uncertain ROI of high-cost trials and long drug development timelines. However, development costs can be improved with faster, predictive, and scalable discovery approaches. While pharma companies often have rich internal datasets, the data may not provide adequate coverage for each discovery challenge and datasets may also not be curated and balanced for training AI models. Data availability and curation remain a challenge across the industry.
Given the scale and complexity of today’s sequence space, accelerating antibody innovation requires methods that don’t sacrifice scientific rigor but preserve speed.
SandboxAQ™ is streamlining the antibody design process with its Antibody Design Platform. The platform combines protein language models, AI-driven structure prediction, deep learning–guided side-chain refinement, and enhanced alchemical sampling to enable fast, cost-efficient affinity predictions—without requiring crystal structures or high-end compute resources. The platform reduces experimental load, enables earlier candidate triage, and supports scalable optimization and design pipelines. By simulating affinity and developability in silico, teams can deprioritize lower-performing variants before committing to in vitro or in vivo testing—ultimately accelerating the design cycle while minimizing unnecessary animal studies.
SandboxAQ’s physics-based, AI-powered in silico workflow addresses each of these challenges with novel computational algorithms that scale and perform well under constraints. As part of SandboxAQ’s next-generation AQBioSim platform, each module works together to address key barriers in antibody design through a physics-grounded, AI-accelerated framework:
Together, these capabilities help antibody discovery teams accelerate candidate refinement, reduce dependency on wet-lab cycles, and improve confidence in early decision-making, allowing R&D leaders to scale programs with fewer experimental bottlenecks. SandboxAQ is bridging the gap between conventional wet-lab experimentation and computational chemistry, supporting both discovery and translational teams with scalable simulation capabilities.
The Antibody Design Platform powers a five-step modular design engine built to accelerate biologics R&D using in silico methods. This approach is central to SandboxAQ’s Large Quantitative Model(LQM) initiative, which merges physics and AI to enhance prediction fidelity with reduced dependence on experimental training data.
The end-to-end pipeline supports drug development software and computational chemistry workflows with scalable, high-fidelity predictions—enabling antibody design in approximately two weeks per computational cycle. By simulating affinity in silico, AQFEP reduces the need for antibody engineering in animal models, which streamlines development and supports ethical R&D practices.
To benchmark AQFEP’s performance, our team at SandboxAQ applied the platform to the 1BJ1 Fab–antigen system which is a well-studied antibody–antigen complex with a canonical binding interface. This validation and methods development effort formed the basis of our presentation at the Festival of Biologics USA 2025 and poster presentation at the 2025 Workshop on Free Energy Methods in Drug Design.
A set of 23 mutation variants (single and multi-point) were used to evaluate performance across different model conditions. Highlights from our validation study include:
These results position AQFEP as a scientifically validated platform for pharmacological modeling and antibody optimization, particularly in low-data or structure-free scenarios.
[Download the Poster]: From Cofolding to FEP – Unveiling the Path to Absolute Antibody Affinities
The AQBioSim team continues to invest in expanded capabilities, usability, and performance across the platform. Upcoming enhancements include:
These updates further position AQBioSim as a comprehensive drug discovery and development software platform for scalable, simulation-driven biologics R&D.
R&D organizations need more than just better software—they need scientific partners who understand the complexity of therapeutic discovery and can scale innovation across programs. Our team at SandboxAQ is excited to tackle challenging problems in the field to deliver innovation that accelerates your discovery processes. Our milestone-driven approach ensures alignment on scientific priorities, shared accountability, and tangible outcomes at each stage of the collaboration.
Explore how our Antibody Design Platform can support your next antibody campaign—connect with our scientific team to start the conversation.
Mary Pitman, PhD, is a Staff Research Scientist at SandboxAQ specializing in combining AI with physics-based approaches. Dr. Pitman leads the Drug Discovery methods development team to develop scientific software for improved therapeutic outcomes and biological insights. Her research focuses on biophysics, graph theory, and free energy perturbation. She presented this work at the Festival of Biologics USA 2025.
