Enhancing Safe and Controllable Protein Generation via Knowledge Preference Optimization

2025/07/15 by Yuhao Wang, Wang, Yuhao, Keyan Ding +13 · 1 citation
Biochemistry, Genetics and Molecular Biology · Computer Science · #Artificial Intelligence (cs.AI) #Computational Drug Discovery Methods #FOS: Computer and information sciences #Genetics, Bioinformatics, and Biomedical Research #Viral Infectious Diseases and Gene Expression in Insects

paper · pdf · doi:10.48550/arxiv.2507.10923

openalex publication_date 2025/07/15 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Protein language models have emerged as powerful tools for sequence generation, offering substantial advantages in functional optimization and denovo design. However, these models also present significant risks of generating harmful protein sequences, such as those that enhance viral transmissibility or evade immune responses. These concerns underscore critical biosafety and ethical challenges. To address these issues, we propose a Knowledge-guided Preference Optimization (KPO) framework that integrates prior knowledge via a Protein Safety Knowledge Graph. This framework utilizes an efficient graph pruning strategy to identify preferred sequences and employs reinforcement learning to minimize the risk of generating harmful proteins. Experimental results demonstrate that KPO effectively reduces the likelihood of producing hazardous sequences while maintaining high functionality, offering a robust safety assurance framework for applying generative models in biotechnology.

Citations

• In-Context Unlearning: Language Models as Few Shot Unlearners
• InstructProtein: Aligning Human and Protein Language via Knowledge Instruction
• Direct Preference Optimization: Your Language Model is Secretly a Reward Model
• RRHF: Rank Responses to Align Language Models with Human Feedback without tears
• Large language models generate functional protein sequences across diverse families
• Knowledge Unlearning for Mitigating Privacy Risks in Language Models
• ColabFold: making protein folding accessible to all
• Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
• Training language models to follow instructions with human feedback
• OntoProtein: Protein Pretraining With Gene Ontology Embedding
• Highly accurate protein structure prediction with AlphaFold
• Pfam: The protein families database in 2021
• BERTology Meets Biology: Interpreting Attention in Protein Language Models
• Machine Unlearning
• RLCD: Reinforcement Learning from Contrastive Distillation for Language Model Alignment

Cited by

• SafeProtein: Red-Teaming Framework and Benchmark for Protein Foundation Models

Related

• Reinforcement Learning for Sequence Design Leveraging Protein Language Models
• Controllable Protein Sequence Generation with LLM Preference Optimization
• Towards Responsible Governance of Biological Design Tools
• g-DPO: Scalable Preference Optimization for Protein Language Models
• Preference optimization of protein language models as a multi-objective binder design paradigm