Updated for 2026: A revised 2026 beginner guide to ESM3, Biohub, first experiments and basic scientific validation.
Start with a question
Decide whether you want to generate a candidate protein, encode protein sequences, predict a 3D structure or explore a database. ESM3 suits multimodal generation; ESMC, ESMFold2 and Atlas address different questions.
Set up a controlled environment
Use a supported Python version in a virtual environment and check the maintained Biohub ESM repository for installation instructions. A common starting point is python -m pip install esm. Confirm current API client and model IDs in Biohub documentation. Do not use an ESM2 checkpoint as an ESM3 demonstration.
Begin with documented examples
Follow an official quickstart using a harmless toy sequence or published educational example. Keep keys in environment variables, never inside notebooks or browser code. Note whether inference runs locally or on a remote API, as limits and data policies differ.
Interpret results correctly
Output candidates are hypotheses. Compare to baselines, evaluate uncertainty and consider prospective validation where appropriate. Record the package version, checkpoint identifier, prompts, seeds, filters and execution environment.
Evidence and further reading
Primary sources: Biohub ESM3 model, Biohub maintained ESM repository, and Biohub learning resources. Check model cards and release notes for version-specific instructions.
Editorial note: Updated October 2026. Distinguish scientific literature and current product documentation from speculative applications; verify current API details before running code.