A LinkedIn post from SandboxAQ highlights the company’s presence at ICLR, where it is showcasing three research posters focused on physics‑informed AI for drug discovery. The post outlines work on SAIR, a synthetic structural dataset of more than 5 million protein–ligand 3D complexes designed to support deep learning models for protein‑ligand interactions and binding‑affinity prediction.
Claim 55% Off TipRanks
- Unlock hedge fund-level data and powerful investing tools for smarter, sharper decisions
- Discover top-performing stock ideas and upgrade to a portfolio of market leaders with Smart Investor Picks
According to the post, this dataset is already being used to power SandboxAQ’s internal experimental and synthetic binding‑affinity models and could enable structure‑based machine learning that generalizes better in low‑data drug discovery settings. Additional work presented at the GemBio workshop reportedly combines experimental data with roughly 80,000 physics‑based AFEP calculations to improve model accuracy and hit rates, particularly in out‑of‑distribution scenarios.
The LinkedIn post also describes ALPAQAFlow, a selectivity‑aware generative model aimed at designing potent, selective, and synthesizable small molecules within realistic design–make–test cycles. In simulated campaigns across challenging targets, the model is said to outperform exhaustive screening of large purchasable libraries on potency and selectivity, and to uniquely generate new chemical series meeting stringent multi‑target criteria.
For investors, the research emphasis suggested in the post points to continued investment in AI‑driven drug discovery infrastructure rather than immediate revenue events. If these approaches translate from simulations and conference demonstrations into adoptable tools for pharmaceutical and biotech partners, SandboxAQ could strengthen its competitive position in AI‑enabled drug discovery and expand its addressable market in licensing, collaborations, and SaaS‑style offerings over the medium term.