Unravelling cyclic peptide membrane permeability prediction: a study on data augmentation, architecture choices, and representation schemes

2025/01/01 by Alfonso Cabezón, Erik Otović, Daniela Kalafatović +3 · 1 voice
Biochemistry, Genetics and Molecular Biology · Immunology and Microbiology · #Machine Learning in Bioinformatics #Chemical Synthesis and Analysis #Antimicrobial Peptides and Activities

paper · pdf · doi:10.1039/d4dd00375f

openalex publication_date 2025/01/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/07

Abstract

A machine learning approach integrating cyclic structure modeling and data augmentation improves cyclic peptide membrane permeability prediction. The best model is deployed in CYCLOPS, a web tool for rapid permeability assessment in drug discovery.

Citations

• Fundamentals of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) network
• Therapeutic peptides: current applications and future directions
• Peptide-based therapeutics: challenges and solutions

Discussions

• In our next featured article, Cabezón, Otović, Mausa et al. discuss the research in to ML models for cyclic peptides which led to their web-based permeability tool, CYCLOPS. Read the paper here: doi.o [bsky, 1 points, 0 comments]

Related

• A decoupled alignment kernel for peptide membrane permeability predictions
• LengthLogD: A Length-Stratified Ensemble Framework for Enhanced Peptide Lipophilicity Prediction via Multi-Scale Feature Integration
• StaPep: an open-source tool for the structure prediction and feature extraction of hydrocarbon-stapled peptides
• Drug-membrane permeability across chemical space
• PepDoRA: A Unified Peptide Language Model via Weight-Decomposed Low-Rank Adaptation