CPP2Vec: A representation learning approach for cell-penetrating peptides prediction

Background: Cell-penetrating peptides (CPPs) facilitate the delivery of a variety of therapeutic molecules across the plasma membrane, from small chemical substances to nucleic acid-based macromolecules, such as antisense oligonucleotides (ASOs). Among neutral ASOs, peptide nucleic acids (PNAs) and phosphorodiamidate morpholino oligomers (PMOs) have been extensively studied as potential medical treatments for Duchenne Muscular Dystrophy (DMD), a severe genetic disease that causes muscle degeneration progressively. Over the last few decades, many in silico methods have emerged to detect novel CPPs, counterbalancing the cost of wet-lab experiments. Results: In this study, we propose CPP2Vec, a Word2Vec-based CPP prediction method, where the Word2Vec technique is used to represent amino acid sequences of peptides. To address the limited sequence diversity, sparse biological grounding, and the still poorly understood mechanisms underlying CPPs uptake, we constructed CPP2Vec-GenSet, a hybrid dataset that integrates computationally generated peptides with experimentally curated CPPs. This combined resource provides a robust training foundation that supports reliable representation learning and enhances cross-task model performance. Using this framework, we developed three task-specific supervised machine learning models for CPP-Classification, Uptake-Efficiency and PMO-Delivery. The first two models were designed to determine if an unseen peptide is a CPP and to predict its uptake efficiency, respectively, while the PMO-Delivery model predicts whether a peptide could enhance the cellular delivery of a PMO-complex compared to its naked version. Furthermore, we explored an alternative approach using pre-trained protein-based Large Language Models (LLMs) – ProtT5, ProtBERT, and ESM-2 – to generate the embeddings, resulting in three task-specific models, namely CPP2LLM. Benchmarking against state-of-the-art CPP prediction tools demonstrates that CPP2Vec achieves robust predictive performance and generalization across tasks, while maintaining exceptional computational efficiency. Conclusion: In this research, we present a Machine Learning (ML)-based tool that introduces the use of the Word2Vec technique in the field of CPPs prediction. Notably, CPP2Vec automatically learns informative peptide representations directly from sequence data, generalizes reliably across multiple tasks, and achieves high predictive performance with minimal computational resources, providing a reproducible and practical in silico tool to support the early-stage identification and prioritization of CPPs with potential therapeutic relevance. CPP2Vec is available for use at: https://github.com/SSvolou/CPP2Vec. Author summary: Cells are surrounded by a membrane that functions as a highly selective security system, guarding against harmful invaders while maintaining internal stability. However, this same protective barrier also blocks many beneficial therapeutic molecules from reaching their intracellular targets, so even promising drugs may fail simply because they cannot cross it.