PepTune: De Novo Generation of Therapeutic Peptides with Multi-Objective-Guided Discrete Diffusion

We present PepTune, a multi-objective discrete diffusion model for simultaneous generation and optimization of therapeutic peptide SMILES. Built on the Masked Discrete Language Model (MDLM) framework, PepTune ensures valid peptide structures with a novel bond-dependent masking schedule and invalid loss function. To guide the diffusion process, we introduce Monte Carlo Tree Guidance (MCTG), an inference-time multi-objective guidance algorithm that balances exploration and exploitation to iteratively refine Pareto-optimal sequences. MCTG integrates classifier-based rewards with search-tree expansion, overcoming gradient estimation challenges and data sparsity. Using PepTune, we generate diverse, chemically-modified peptides simultaneously optimized for multiple therapeutic properties, including target binding affinity, membrane permeability, solubility, hemolysis, and non-fouling for various disease-relevant targets. In total, our results demonstrate that MCTG for masked discrete diffusion is a powerful and modular approach for multi-objective sequence design in discrete state spaces.

Peptides — short protein-like drugs such as GLP-1 receptor agonists (think Ozempic and Wegovy) — have transformed treatment for diabetes and obesity. But what if we could create Ozempic-like therapies to treat cancer, reproductive, or neurodegenerative disease? PepTune moves us toward that goal. Unlike traditional methods that depend on a protein’s 3D structure, PepTune works directly from the target sequence, enabling peptide design for disease-driving proteins that are often flexible and disordered. Our model generates entirely new peptides that not only bind their targets but also meet multiple drug-like criteria — such as solubility, low toxicity, and membrane permeability — critical for clinical success. At the heart of PepTune is a powerful multi-objective algorithm that combines a generative model with a search strategy to learn which sequences optimally balance these competing properties. The result: peptide drugs that are not only effective, but also developable, allowing us to make better, safer therapies faster.