PDB to FASTA
Extract protein, DNA, or RNA sequences from a PDB or mmCIF structure. Choose the sequence source before exporting individual chains.
Structure contents stay in your browser. Plain-text files up to 10 MiB.
Structure input
Paste text, drop a file, or start from an example.
FASTA output
Load a structure, choose a source, then extract its sequences.
Choose the sequence that fits your analysis
Declared sequence
PDB SEQRES records and mmCIF entity_poly_seq describe the polymer in the deposited structure, including residues without coordinates. This may differ from the full biological protein or a reference database sequence.
Observed sequence
Coordinate records describe residues present in the first model. Alternate atom locations count once. Gaps are omitted, so this sequence does not restore missing residues or preserve alignment positions.
Chain identifiers
Each selected chain becomes a FASTA record. For mmCIF, label_asym_id identifies chains; author chain IDs appear alongside them when available. FASTA lines wrap at 60 characters.
A missing residue changes the result
Both examples declare AGSMX for the protein chain and ACG for the DNA chain. Serine has no coordinate record, so observed mode gives AGMX and ACG. The repeated alanine atom counts once. The mmCIF example labels the protein chain AA, with author ID A.
Protein chain
DNA chain
No coordinates
Related sequence tools
Format references
PDB to FASTA questions
Can I convert mmCIF to FASTA?
Yes. Upload a plain-text .cif or .mmcif file with one data block. Declared mode requires entity_poly_seq plus struct_asym. Observed mode uses atom_site. BinaryCIF and compressed files need to be converted or decompressed first.
Why is the observed sequence shorter?
Residues without coordinates are absent from observed output. Choose declared mode if you need the deposited polymer sequence. If sequence records are missing, the tool reports that limitation instead of guessing the missing residues.
How are modified residues, ligands, and multiple models handled?
Standard amino acids and nucleotides use one-letter codes. Modifications use the standard parent supplied by MODRES or chem_comp when available; MSE maps to M. Unrecognized or conflicting residues become X. PDB HETATM records are included when they match SEQRES residues or recognized modifications; other unknown HETATM residues are excluded. mmCIF polymer residues use label_seq_id. Other ligands and solvent are excluded. Only the first coordinate model is exported.
Does this align sequences or predict a structure?
No. This extracts sequence records already in a structure file. FASTA output does not include atom coordinates, residue numbering, insertion codes, or an alignment.