PTM-dCN: Latent Space Control for Post-Translational Modification–Aware Protein Design

Abstract Post-translational modifications (PTMs) are critical for protein function, yet their precise design by harnessing site specific information derived from native proteins remains challenging. Here, we present a deep learning–based PTM design framework that integrates latent diffusion models with ControlNet for sequence generation with site-specific PTM-control. The framework incorporates a PTM-aware protein language model featuring extractor, trained on a curated SwissProt PTM dataset with specialized modification tokens. Through de novo generation of protein sequences with designated PTM sites, our framework facilitates the exploration of PTM-driven functional landscapes and advances position-aware protein engineering. Competing Interest Statement The authors have declared no competing interest. Footnotes {zhangsitao{at}sjtu.edu.cn,ruiqing.br{at}sjtu.edu.cn} edward77{at}mit.edu 178069354{at}sjtu.edu.cn