Chemically modified peptides are promising therapeutic scaffolds, but their database annotations and sequential labeling or adding metadata to raw data, so a machine learning model can learn from it remains difficult. Many peptides consist of features like non-proteinogenic residues, terminal derivatization, macrocyclization, and covalent links between chains. Chemical structure annotations of such modified peptides are only partly covered by existing sequence-to-structure tools.
We developed a Python application PepSMILES[1] that constructs atom-resolved molecular graphs from structured peptide records and exports stereochemical SMILES, V2000 MOL files, and two-dimensional images. The application currently includes more than 650 unusual-amino-acid blocks, together with 370 curated N- and C-terminal modifications. It supports up to 50 intrachain annotations found in the current DBAASP export and nine interchain annotations, including disulfide, amide, ester, thioether, carbon–carbon, heterocyclic, and multibridge closures. Assembly is metadata-driven: each residue and terminal template identifies the atoms involved in peptide or cross-link formation and any leaving groups, while chemistry-specific rules apply the required bond changes for each intra- or interchain modification. When applied at database scale, the workflow returned SMILES for 25,243 of 25,306 eligible DBAASP[2] records.
Applications like p2smi[3], cyclicpeptide[4], and PepLink[5], none reports this same combination of unusual-residue coverage, terminal derivatization, chemically diverse cyclization, and covalent multimer assembly.
By reducing manual structure drawing and taking advantage of reusable chemical rules, the application can shorten dataset preparation, improve structural consistency, and allow experimentally described peptides to be compared directly with small molecules in established drug-discovery pipelines.
This approach makes chemically complex peptides accessible to cheminformatics, QSAR, virtual-screening, and machine-learning workflows.
1. https://dbaasp-dev.ibrsp.org/tools?page=smiles-generator
2. https://dbaasp.org/home
3. https://github.com/AaronFeller/p2smi
4. https://github.com/dfwlab/cyclicpeptide
5. https://github.com/DragonDescentZerotsu/PepLink