Morphology-Specific Peptide Discovery via Masked Conditional Generative Modeling

2025-09-02

AI Summary

This study addresses the challenge of navigating the vast peptide sequence space to enable targeted generation of peptides with predefined self-assembly morphologies—specifically fibrillar or spherical structures. We propose PepMorph, an end-to-end generative framework that achieves morphology-controllable peptide design for the first time. Methodologically, PepMorph employs a Transformer-based conditional variational autoencoder (cVAE) augmented with a masking mechanism to enhance sequence controllability under morphological constraints. To ensure physical plausibility, we integrate coarse-grained molecular dynamics (CG-MD) simulations for structural validation and construct a novel training dataset enriched with physics-informed chemical descriptors. Experimental results demonstrate that PepMorph achieves 83% accuracy in generating peptides with target morphologies; moreover, it successfully produces novel, experimentally verifiable peptide sequences exhibiting bona fide self-assembly activity. This work establishes a scalable, rational generative paradigm for designing functional peptide materials.

Technology Category

• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Peptide self-assembly prediction offers a powerful bottom-up strategy for designing biocompatible, low-toxicity materials for large-scale synthesis in a broad range of biomedical and energy applications. However, screening the vast sequence space for categorization of aggregate morphology remains intractable. We introduce PepMorph, an end-to-end peptide discovery pipeline that generates novel sequences that are not only prone to aggregate but self-assemble into a specified fibrillar or spherical morphology. We compiled a new dataset by leveraging existing aggregation propensity datasets and extracting geometric and physicochemical isolated peptide descriptors that act as proxies for aggregate morphology. This dataset is then used to train a Transformer-based Conditional Variational Autoencoder with a masking mechanism, which generates novel peptides under arbitrary conditioning. After filtering to ensure design specifications and validation of generated sequences through coarse-grained molecular dynamics simulations, PepMorph yielded 83% accuracy in intended morphology generation, showcasing its promise as a framework for application-driven peptide discovery.

Problem

• Screening vast peptide space for morphology categorization
• Generating novel sequences for specified fibrillar or spherical assembly
• Predicting peptide self-assembly for biomedical material design

Innovation

• Transformer-based Conditional Variational Autoencoder
• Masking mechanism for peptide generation
• Coarse-grained molecular dynamics validation