Deep Origin announced a comprehensive preprint on bioRxiv describing a major advance in virtual screening, validated prospectively in physical wet-lab assays, discovering chemically novel small molecules against four challenging therapeutic targets. The novel framework, DODock and DOScore, maintains predictive accuracy on novel biological targets where conventional machine-learning models collapse. On the Runs N' Poses benchmark, DODock's pose prediction success rate exceeds that of all models across nearly all similarity categories. On CASF-2016 (redocking), DODock outperforms Glide, LeDock, Vina, and AutoDock on both the 2 Å and stricter 1 Å criterion. On the OpenBind benchmark, DODock performs best of all methods tested, at 80% on the harder criterion (81% on the easier one), also beating GNINA (66%) and SMINA (46%). Across protein families, DODock's 2 Å success rates range from about 76% to 93% with a mean of 82%. In a blind test, Deep Origin’s model placed AstraZeneca's oral PCSK9 inhibitor at an unexpected pocket months before a crystal structure confirmed the pose to ~1.2 Å. A prospective screen across 80 billion virtual compounds yielded a 30.6% hit rate on CD73, ~100x higher than a prior machine-learning screen. By pairing machine learning with physics, DODock maintains over 50% pose accuracy on novel