Predicting a protein’s shape from its sequence

The human problem

Experimental structure determination cannot keep pace with the vast number of protein sequences found in nature.

The prior constraint

Structure prediction and experimental measurement both demanded substantial time and compute per protein.

AI’s actual role

Learned sequence representations are converted into predicted atomic structure without requiring a multiple-sequence alignment for every target.

The documented result

The Science study used the approach to help create the ESM Metagenomic Atlas, a public collection of predicted structures for hundreds of millions of metagenomic proteins.

Why it may matter

Portable predictions can help researchers choose where to investigate, but a model output is evidence to test rather than a measured structure.

Limitations

Predicted structures vary in confidence and require scientific follow-up. The MIT repository and CC-BY-4.0 Atlas do not amount to a complete released training corpus and recipe.

The peer-reviewed paper and public artifacts establish the research result; neither establishes experimental validity for every predicted protein.

Unresolved questions