Current tools excel at the static binding problem. To advance to creating true therapeutics, models must incorporate the physics of displacing solvent and ions from an interface—currently neglected but one of the "biggest enemies" of strong binding in a physiological context.
The concept of an "undruggable" target is a misnomer, according to Pacesa. Any failure to create a binder for a specific protein site is a limitation of the current design method or modality, not an intrinsic property of the target. He posits that, with the right approach, a binder can be designed for any site.
Unlike generative tools that always produce an output, BindCraft sometimes yields no passing designs. This "failure" is a valuable feature, acting as a strong negative predictor that saves researchers months of wasted lab effort on low-probability targets. This builds user confidence in the designs that do pass.
The design tool isn't a passive executor. Its multi-component loss function, optimizing for properties like foldedness, can override a user's chosen binding site if it's suboptimal. This "AI agency" is a key feature that contributes to its high success rate in the lab.
The initial motivation for BindCraft wasn't just to design better proteins, but to avoid the laborious, low-yield process of yeast display screening. This personal frustration with an inefficient workflow drove the development of a computational tool that dramatically increased hit rates from 1-in-1000 to 7-in-10.
An anecdote about a "wonky" BindCraft design with disconnected beta sheets, which experts predicted would fail, highlights a key trend. The resulting binder was one of the best ever produced, suggesting AI models are extracting structural principles that go beyond traditional human "protein literacy" and intuition.
The pipeline's high success rate stems from its final filter, which uses an AlphaFold model trained only on single proteins (monomers) to predict a protein complex. The rationale is that if a model naive to complexes can still predict the interaction, the interface must be exceptionally strong and well-defined.
Pacesa argues that closed-source models won't significantly outperform open-source tools because most rely on the same public PDB data. The true competitive advantage lies not in tweaking algorithms but in generating massive, proprietary, high-quality experimental datasets that can train genuinely superior models.
Unlike antibodies with flexible loops allowing for induced fit, de novo designed proteins are hyper-rigid. This pre-organized structure leads to rapid binding (fast on-rates) but also makes them susceptible to rapid unbinding (fast off-rates), as the rigid interface is more easily displaced by solvent.