EvolutionaryScale (ESM3)

Identify target substrate structure.

Prompt ESM3 with active site geometry.

Generate 1,000 candidate sequences.

Filter by pLDDT structure confidence.

Synthesize top 5 candidates for wet-lab testing.

Input existing antibody sequence and structure.

Mask the CDR loops.

Sample new residues in the masked region using ESM3.

Rank by log-likelihood of binding conformation.

Select candidates with highest evolutionary fitness.

Input the native protein structure.

Run a 'fitness landscape' scan for all possible single-point mutations.

Identify mutations that increase structural rigidity.

Combine top mutations using the model's joint probability estimation.

Validate structure via ESMFold.

Feed current viral protein sequences into ESM3.

Generate probable mutation clusters based on evolutionary trajectory.

Analyze structural changes in binding epitopes.

Design preventative vaccine sequences.

Simulate antibody evasion.

Define the antigen's structural coordinates.

Prompt ESM3 to 'build a scaffold' around the antigen.

Iterate to ensure the scaffold is self-assembling.

Check for low immunogenicity of the scaffold sequence.

Export PDB for synthesis.

Batch process unknown sequences through ESM3.

Compare internal representation clusters with known functional families.

Assign high-confidence functional labels based on structural similarity.

Identify novel metabolic pathways.

Catalog findings in a searchable database.

Specify the chemical ligand binding pocket.

Use ESM3 to design a responsive 'switch' mechanism.

Optimize for signal-to-noise ratio in the generated structure.

Verify folding in a virtual cell environment.

Predict expression levels in E. coli.