New Model Predicts Complex Lasso Peptide Structures
The framework successfully models threaded peptide topologies that previously challenged general-purpose prediction software.
Updated on Oct. 16, 2026 in Biotech
Live Poll
Do you believe that artificial intelligence will successfully accelerate the discovery of new life-saving medicines?
Researchers have released HighFold-Lasso, a new computational tool designed to predict the structures of complex lasso peptides. This research-stage framework achieved a 100% success rate in correctly predicting the topology for 52 experimentally determined test structures.
Why it matters
Lasso peptides, defined by their threaded structures and isopeptide linkages, have historically defied accurate modeling by general-purpose structure prediction tools. This framework addresses these specific topological challenges, enabling more reliable structural insights for these bioactive molecules.
The system models the four classical classes of lasso peptides with a mean RMSD C-alpha—a measure of atomic position deviation from a reference—of 1.942 Angstrom for ring-loop regions and 2.375 Angstrom for full peptides.
The details
HighFold-Lasso utilizes custom covalent connectivity and lasso scaffold-guided adaptive low-noise initialization to navigate the unique constraints of these peptides. It employs conditional structure generation, which incorporates lasso conformational priors—mathematical assumptions about the peptide's likely folded state—to handle threaded topologies and noncanonical amino acids.
Timeline
October 16, 2026: The research article was published.
The Tech Race
The development marks a departure from general-purpose structure prediction models, which often fail to account for the unique threading and disulfide connectivity of lasso peptides. By introducing domain-specific scaffold guidance, the framework provides a specialized alternative to the broader, non-specialized protein folding pipelines that currently dominate the field.
This tool is currently a research-stage instrument intended for bioinformaticians and drug discovery teams modeling peptide-based therapeutics. Users can integrate it into workflows involving noncanonical amino acids and complex cyclic peptides to improve structural accuracy in protein design.
The takeaway
This framework offers a specialized solution for structural biologists working with threaded peptides that exceed the capabilities of off-the-shelf folding software. Researchers should track future benchmarking updates to see if the model maintains this accuracy on peptide classes outside the current set.
Further reading
For broader developments in molecular modeling, explore the Biotech section.
More information
Review the technical findings in the HighFold-Lasso research article.
Source note: This article includes information reported by Biorxiv.
Live Poll
Do you believe that artificial intelligence will successfully accelerate the discovery of new life-saving medicines?
