AlphaFold
Predicts protein 3D structure from amino acid sequence with near-experimental accuracy. AlphaFold3 extends this to predict interactions with DNA, RNA, and small molecules.
What it does
AlphaFold, developed by DeepMind, predicts a protein’s 3D structure from its amino acid sequence. The AlphaFold Protein Structure Database now hosts precomputed predictions for virtually all known proteins — over 200 million structures across nearly every known organism. AlphaFold3 extends this to predict interactions between proteins, DNA, RNA, and small molecules, enabling applications in drug design and understanding biological mechanisms.
This is a research model and database, not a SaaS product. There’s no “sign up” button — access is through the database for precomputed predictions, or the associated code repositories for custom predictions.
Best for
Any researcher in structural biology, biochemistry, or structure-based drug design who needs a structure prediction without running their own crystallography or cryo-EM. AlphaFold turned a problem that once took years of experimental work into a database lookup — which is why it’s arguably the most scientifically significant tool in this directory, and the default starting point for nearly any structural biology question.
Used by more than three million researchers across 190 countries.
Pricing
Free. The AlphaFold Database is freely accessible for precomputed structures. The associated code is open source. AlphaFold3 has an online server for limited predictions.
Strengths
- Free, openly accessible database of 200M+ precomputed protein structures
- Near-experimental accuracy for many structure types — competitive with X-ray crystallography in well-benchmarked cases
- AlphaFold3 adds interaction prediction for the full range of biologically relevant molecules
- Established scientific track record — results widely validated by experimental labs
- Open source: you can run custom predictions on your own hardware if you need beyond what the database holds
Limitations
- Predicts structure, not function with certainty — a confident structure prediction doesn’t guarantee correct biological behavior; experimental validation is still required
- Performance on highly disordered regions and very large complexes is more variable
- AlphaFold3’s full capabilities are primarily available through the limited online server, not fully open-source yet (unlike AlphaFold2)
- Isomorphic Labs’ newer proprietary engines (not independently accessible) suggest the state-of-the-art has moved further — but AlphaFold remains the open benchmark
How it compares
| vs. | Key difference |
|---|---|
| RFdiffusion / ProteinMPNN | AlphaFold predicts structure of existing sequences; RFdiffusion/ProteinMPNN design new protein sequences from scratch |
| RoseTTAFold | Both are structure predictors; AlphaFold is more widely validated and has a larger precomputed database |
| Cryo-EM / X-ray crystallography | Experimental methods remain the gold standard; AlphaFold predictions are often used to guide or interpret experimental results |