High-Speed Protein Structural Analysis with Foldseek is Now Available on Vecura

What is Foldseek?

Foldseek is an ultra-fast structural alignment toolkit that revolutionizes protein structure comparison by representing 3D protein tertiary contacts as a compressed 20-letter "3Di" structural alphabet. By leveraging the highly optimized MMseqs2 sequence-search machinery, it enables structural searches to be performed up to 1000x faster than traditional methods like DALI or TM-align. It is especially useful for researchers needing to search or cluster millions of protein structures or complexes without the prohibitive computational costs of standard all-atom alignment techniques.

What can users do with Foldseek on Vecura?

With Foldseek on Vecura, users can:

• Search Single-Chain Structures: Quickly query novel protein structures against vast databases like the PDB or AlphaFold Database to identify structural homologs.
• Cluster Large Datasets: Effortlessly perform large-scale clustering of structural predictions to identify representative sets and reduce redundancy in protein collections.
• Analyze Protein Complexes: Perform advanced multi-chain structural searches and clustering using Foldseek-Multimer to compare interface geometry and complex-level architecture.
• Sequence-Based Searching: Utilize ProstT5 integration to search structure databases using only FASTA sequences, bypassing the need for pre-computed 3D coordinates.

What the output means

The output provides comprehensive structural alignment reports, including E-values, bit scores, TM-scores, and interface lDDT metrics. These metrics provide quantitative insights into the structural similarity between proteins or complexes.

This output should be used to support scientific decision-making, such as identifying evolutionary relationships, inferring protein functions, or validating structural models. It does not replace experimental validation.

Why this matters

The rapid expansion of the structural biology landscape, driven by high-throughput protein folding predictions, has created a critical bottleneck in how we compare and classify these vast numbers of structures. Traditional, slower alignment methods struggle to scale to the millions of entries now available.

Foldseek bridges this gap by providing near-identical sensitivity to exhaustive methods at a fraction of the time, democratizing access to large-scale structural analysis. By enabling the rapid exploration of the "protein universe," it empowers researchers to discover remote evolutionary connections and structural patterns that were previously computationally inaccessible.

• Developed by: Steinegger Lab
• Source: Official GitHub Repository
• Reference: van Kempen et al., Nature Biotechnology (2023); Barrio-Hernandez et al., Nature (2023); Kim et al., Nature Methods (2025).