AQBioSim

AQBioSim, developed by SandboxAQ, is a comprehensive end-to-end drug discovery and development platform powered by Large Quantitative Models (LQMs) derived from AI and physics-based methods. Designed for pharmaceutical and biotech organizations working on small molecules and biologics, AQBioSim accelerates every stage of the drug discovery lifecycle — from early-stage target identification through to clinical candidate selection — delivering faster and more accurate solutions to the most challenging problems in the field.

SandboxAQ acts as a strategic partner, integrating seamlessly into customer programs to enhance the generation of novel molecular drug IP and clinical assets. Through a milestone-based project approach, SandboxAQ shares in the risk and incentives, ensuring that LQM solutions are tailored to specific discovery and development needs while remaining broadly applicable and reusable across future projects for long-term value.

Core Platform Capabilities

• Precision in Small Molecule and Biologics Discovery: Molecular interactions and behaviors are simulated to optimize critical factors including potency, efficacy, and safety across both small molecule and biologic modalities.
• Seamless Drug Discovery Process: Comprehensive support spans target discovery, hit identification, lead optimization, and toxicity prediction, delivering actionable insights that enable faster and smarter decision-making at every stage.
• LQM-Driven Acceleration: Large Quantitative Models rapidly analyze multi-modal data to identify the most promising targets and candidates, significantly compressing drug development timelines.

IDOLpro — Generative Chemistry AI

• IDOLpro is a guided generative chemistry AI built to design novel, optimized 3D drug-like molecules in minutes on a single GPU.
• Trained on a combination of public data and physics-based simulations, IDOLpro creates molecules with high binding affinity to target proteins and optimized synthetic accessibility.
• The model leverages AWS infrastructure, combining diffusion models with multi-objective optimization to guide molecule generation.
• Benchmarked against CrossDocked and Binding MOAD datasets, IDOLpro achieves 3.4x better binding affinity compared to state-of-the-art methods and has generated compounds that outperform experimentally validated molecules.
• The approach compresses months of traditional drug discovery work, underscoring the transformative potential of cloud-based, guided AI for accelerated drug design.

AQFEP — Absolute Free Energy Perturbation

• AQFEP is an innovative absolute free energy perturbation solution that enables rapid and cost-effective exploration of enormous chemical spaces to identify promising drug candidates for experimental investigation.
• Unlike traditional relative free energy methods, AQFEP requires no reference molecule, unlocking the ability to discover first-in-class molecules through an integrated workflow.
• The solution efficiently screens large and diverse chemical libraries virtually, combining active learning with a rigorous physics-based scoring function.
• Performance has been validated in ranking structurally related ligands, virtual screening hit rate enrichment, and active learning chemical space exploration.
• SandboxAQ has disclosed the largest reported collection of free energy simulations to date, further demonstrating the power of this approach.
• AQFEP has shown exceptional results in historically "undruggable" targets, particularly in neurodegeneration and oncology, for both hit identification and lead optimization.

SAIR Dataset and OpenFold Ecosystem

• SandboxAQ, in collaboration with the OpenFold Consortium, has launched AQAffinity — an open-source, structure-free binding affinity prediction tool built on OpenFold3.
• OpenFold3 is a fully open-source frontier AI platform for structure-aware drug discovery.
• The SAIR dataset is the largest publicly available binding affinity dataset with cofolded 3D structures, providing a significant resource for the broader drug discovery research community.

Demonstrated Impact and Use Cases

• In collaboration with UCSF, AQBioSim expanded chemical exploration space from 250,000 molecules to 5.6 million, identifying candidate molecules with a hit rate 30 times greater than conventional approaches, as reported by Nobel Laureate Dr. Stanley Prusiner.
• The platform has enabled medicinal chemistry on multiple promising starting points for previously undruggable protein targets, as validated in work with Riboscience.
• Case studies include partnerships with iOncologi Cancer Therapeutics and UCSF Drug Discovery, with additional collaborations involving Sanofi, Pfizer, the Michael J. Fox Foundation for Parkinson's Research, and Columbia University.
• Applications span oncology, neurodegeneration (including Alzheimer's and Parkinson's disease), and other major therapeutic areas.

AQBioSim is backed by $300 million in funding and is deployed in partnership with leading pharmaceutical companies and academic institutions worldwide. The platform leverages cloud infrastructure including AWS, and its open-source contributions through the OpenFold Consortium make frontier AI tools broadly accessible to the drug discovery community.