ActSeek is a fast and accurate active-site search algorithm for the AlphaFold database, and ActSeekN is its web interface for enzyme function prediction from 3D protein structures.
Submit one or more PDB files or UniProt accession numbers and receive functional annotations, including EC numbers and mapped active-site residues.
Reference: Castillo S, Ollila S, ActSeek: Fast and accurate search algorithm of active sites in Alphafold database, Bioinformatics, 2025; btaf424 — https://doi.org/10.1093/bioinformatics/btaf424 . Source code: https://github.com/vttresearch/ActSeek .
Enzyme function is typically described using EC numbers, a four-level hierarchical classification (for example, 1.1.1.1 for alcohol dehydrogenase). Sequence-based methods such as BLAST and HMM profiles work well for close homologs, but can miss distant homologs that share active-site geometry rather than global sequence similarity.
ActSeekN uses partial structural homology of active and binding residues to annotate enzyme function, which is especially useful for novel folds and low-sequence- identity proteins.
A typical protein of 400 to 500 amino acids is processed in about 5 minutes. Jobs are executed in a FIFO queue, and results appear progressively on the query detail page as each structure finishes.
For multi-structure submissions, each structure is dispatched as a separate job and may run in parallel up to available worker capacity.
Resubmitting identical input retrieves cached results immediately without re-queuing. You can also recover the URL from your browser history.
ActSeekN accepts multiple structures per submission. Upload several PDB files or provide UniProt accessions separated by spaces or commas.
For very large batches or automated workflows, use local ActSeek scripts from https://github.com/vttresearch/ActSeek .
Use the UniProt Accession Numbers tab. ActSeekN fetches AlphaFold2- predicted structures from the EBI AlphaFold database. You can also upload your own model as a PDB file.
ActSeekN is freely available for academic and non-commercial use. Commercial users should contact the authors before using ActSeek or ActSeekN in commercial applications.
Castillo S, Ollila S. ActSeek: Fast and accurate search algorithm of active sites in Alphafold database. Bioinformatics, 2025; btaf424. https://doi.org/10.1093/bioinformatics/btaf424
ISMB 2025 poster: https://doi.org/10.7490/f1000research.1120245.1