About LU-Fold

LU-Fold is a Lund University-based facility for helping researchers predict protein structures of interest using the cutting-edge method AlphaFold2 (Nature Methods method of the year, 2021). LU-Fold specialises in high-throughput prediction of protein complexes to predict novel protein-protein interactions. For example, we can predict pairwise interactions of a protein of interest with all other proteins in a proteome to find new binding partners and molecular binding interfaces. We also offer tutorials, workshops and online guides to help others make their own predictions, compare structures with others in the public domain, visualise results and make publication-quality structures.

Services provided

We run as a service, accessing national high performance computing infrastructure to make high-throughput structural predictions. Users do not have to have any previous bioinformatics or structural biology experience.

The services we provide include:

• Predicting pairwise binding interactions of a protein of interest with all other proteins in a proteome
• Predicting the structures of all proteins in a proteome (for instance from a newly sequenced genome)
• Predicting higher order structures of larger complexes
• Predicting the effect of mutations and truncations on interactions
• Workshops on using protein structure prediction tools and protein structure visualisation

User fees

The service is subject to user fees, although initial consultation meetings are free. The fee structure mirrors that used by the National Bioinformatics Infrastructure Sweden (NBIS): 800 SEK / hour of direct work on the user project.

Get started!

Requests for projects (web form)

LU-Fold-related updates from the Atkinson lab, including workshop dates (website for Atkinson lab)

Schematic overview of a typical LU-Fold run

Schematic overview of a typical LU-Fold runIn the illustration/scheme above: The sequence of a protein of interest is co-folded with a proteome of interest (two proteins of the proteome are shown here). Each protein-protein complex is scored with the pDockQ scoring metric and ranked. This provides insights into potential interaction partners and guide future experiment design.