Design of peptides with noncanonical amino acids using flow matching
- 1Department of Molecular Genetics, University of Toronto , Toronto, ON, M5S 3K9,
- 2University of Toronto
- 3Department of Computer Science, University of Toronto , Toronto, ON, M5S 2E4,
Abstract
MOTIVATION: The canonical vocabulary of 20 amino acids limits the chemical space available to proteins and peptides. Expanding this vocabulary to hundreds of noncanonical amino acids (ncAAs) allows the engineering of proteins with novel function and activity, and is of particular interest for therapeutic peptides such as macrocycles, where ncAAs can improve proteolytic stability, membrane permeability, and immunogenicity. However, existing structure-based design tools either cannot model ncAAs at all, or are restricted to a small, fixed vocabulary of ncAAs seen during training, and ncAA data in the Protein Data Bank is scarce and heavily biased. RESULTS: We present NCFlow, a flow matching generative model that places any arbitrary ncAA into a given protein backbone using only its atom types and bond connectivity, and therefore generalizes to ncAAs never seen during training. To supplement sparse training data in the Protein Data Bank, NCFlow is pretrained on millions of small molecule structures and a large set of protein-ligand complexes before finetuning on native noncanonicals found within proteins. NCFlow outperforms AlphaFold3-based methods in the structure prediction of unseen ncAAs. We further present a peptide design pipeline akin to in silico deep mutational scanning, and propose a scoring strategy combining deep learning-based and molecular dynamics-based alchemical binding free energy calculations to identify improved peptide variants. We apply the method on four protein-peptide complex test cases, and observe that incorporating noncanonicals can improve predicted binding affinity by up to -7.0 kcal/mol. AVAILABILITY AND IMPLEMENTATION: NCFlow is freely available at https://github.com/mjslee0921/ncflow.
Related
- Directed Evolution of Threonine Aldolases for the Photobiocatalytic Synthesis of β-Branched-α-Tetrasubstituted and Other Noncanonical Amino Acids
- Beyond Nature's Blueprint: Macrocyclic Peptides Containing Unnatural Amino Acids in Therapeutic Development
- Molecular origami in neuroscience: noncanonical DNA structures and functional nucleic acids in neural regulation and disease
- MONDE·T: A database and interactive webserver for non‐standard amino acids and peptide components in the PDB
- Comprehensive Strategies in the Synthesis of Noncanonical Cyclic Peptides Containing Biaryl-Bridged Linkages
BibTeX
@article{lee2026design,
title = {{Design of peptides with noncanonical amino acids using flow matching}},
author = {Jin Sub Lee and Philip M. Kim},
journal = {Bioinformatics},
year = {2026},
month = aug,
doi = {10.1093/bioinformatics/btag654},
url = {https://doi.org/10.1093/bioinformatics/btag654},
note = {CC BY},
}
RIS
TY - JOUR
TI - Design of peptides with noncanonical amino acids using flow matching
AU - Jin Sub Lee
AU - Philip M. Kim
PY - 2026
DA - 2026/08/31
JO - Bioinformatics
DO - 10.1093/bioinformatics/btag654
UR - https://doi.org/10.1093/bioinformatics/btag654
AB - MOTIVATION: The canonical vocabulary of 20 amino acids limits the chemical space available to proteins and peptides. Expanding this vocabulary to hundreds of noncanonical amino acids (ncAAs) allows the engineering of proteins with novel function and activity, and is of particular interest for therapeutic peptides such as macrocycles, where ncAAs can improve proteolytic stability, membrane permeability, and immunogenicity. However, existing structure-based design tools either cannot model ncAAs at all, or are restricted to a small, fixed vocabulary of ncAAs seen during training, and ncAA data in the Protein Data Bank is scarce and heavily biased. RESULTS: We present NCFlow, a flow matching generative model that places any arbitrary ncAA into a given protein backbone using only its atom types and bond connectivity, and therefore generalizes to ncAAs never seen during training. To supplement sparse training data in the Protein Data Bank, NCFlow is pretrained on millions of small molecule structures and a large set of protein-ligand complexes before finetuning on native noncanonicals found within proteins. NCFlow outperforms AlphaFold3-based methods in the structure prediction of unseen ncAAs. We further present a peptide design pipeline akin to in silico deep mutational scanning, and propose a scoring strategy combining deep learning-based and molecular dynamics-based alchemical binding free energy calculations to identify improved peptide variants. We apply the method on four protein-peptide complex test cases, and observe that incorporating noncanonicals can improve predicted binding affinity by up to -7.0 kcal/mol. AVAILABILITY AND IMPLEMENTATION: NCFlow is freely available at https://github.com/mjslee0921/ncflow.
SN - 1367-4803
SN - 1367-4811
ER -