Research notes from Ligo.
-
Building Artios How we built Artios: atom-level enzyme scaffolding, ligand-aware Potts sequence design, and scalable design pipelines.
-
How Muon Lost Its Geometry How popular Muon implementations lost the scaling rule that made the optimizer geometrically meaningful.
-
The Unreasonable Redundancy of Nature's Protein Folds Why scaling natural sequences into predicted structures may not produce as much new fold diversity as enzyme-design models need.