Frequently asked questions
Can't find an answer? Email hello@foldfunc.xyz.
What sequence format is accepted?+
Single-letter amino acid format only (e.g. MKTIIALSYIFCLVFA...). FASTA headers (lines starting with >) are not supported — paste only the sequence itself. Spaces, newlines, and carriage returns are stripped automatically.
Which amino acids are valid?+
The 20 standard amino acids: A C D E F G H I K L M N P Q R S T V W Y. Non-standard residues (B, J, O, U, X, Z) will return a validation error. Selenocysteine (U) and pyrrolysine (O) are not supported.
What are the sequence length limits?+
Minimum 10 residues, maximum 600 residues. Sequences above 400 residues may be significantly slower on the free ESMFold inference tier — a warning is shown when this applies.
How do I specify mutation positions?+
Enter a comma-separated list of integers using 1-based indexing — the first residue in your sequence is position 1. Example: 14, 35, 52, 101. If left blank, foldfunc selects positions automatically across the sequence.
How accurate is the structure prediction?+
ESMFold achieves accuracy comparable to AlphaFold2 on many single-domain proteins, but can struggle with multi-domain proteins, intrinsically disordered regions, and sequences with little evolutionary data. The pLDDT score is your primary guide to per-region reliability.
What do negative mutation scores mean?+
Mutation scores are log-probabilities from ESM-2. A score of 0 means the wild-type amino acid is exactly as expected at that position. Increasingly negative scores indicate the residue is unusual in evolutionary context — a signal that the position is conserved and likely functionally important.
Why is my analysis taking a long time?+
Structure prediction via ESMFold on the free Hugging Face inference tier can be slow for sequences above 400 residues, or during periods of high demand. Mutation scoring on Modal is typically faster. If analysis takes more than 3–4 minutes, try a shorter sequence or retry during off-peak hours.
Are my sequences stored?+
No. foldfunc does not persist your sequences after returning a result. However, your sequence is passed to third-party APIs (Anthropic, Hugging Face, Modal) as part of the analysis pipeline. See the Privacy Policy for full details.
Can I use foldfunc for clinical or medical decisions?+
No. foldfunc is a research tool. Outputs are probabilistic and generated by machine learning models. They are not validated for clinical use and must not be used to inform medical, diagnostic, or therapeutic decisions.
What is the free tier limit?+
The free tier includes 5 analyses per month. Paid plans with higher limits are coming soon — see the Pricing page for details.
Something went wrong. What should I do?+
If you receive an error, check that your sequence contains only valid amino acids and is within the length limits. For persistent issues, contact us at hello@foldfunc.xyz with a description of what you submitted.