# Enhancing Safe and Controllable Protein Generation via Knowledge Preference Optimization

Source: [Enhancing Safe and Controllable Protein Generation via Knowledge Preference Optimization](<https://www.sota2.com/research/paper/68fa7a101e209131cea80e2c>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-04T04:27:56.578Z

Authors: Yuhao Wang, Keyan Ding, Kehua Feng, Zeyuan Wang, Ming Qin, Xiaotong Li, Qiang Zhang, Huajun Chen

Published: 2025-07-15T02:30:33.000Z

## 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.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Protein Sequence Generation on GB1, PhoQ, UBC9, GFP Evaluation Set](<https://www.sota2.com/research/sota/protein-sequence-generation-on-gb1-phoq-ubc9-gfp-evaluation-set.md>): Protein Sequence Generation; Protein Functional Evaluation GB1, PhoQ, UBC9, GFP; GB1 Score: 0.191.

- [Protein Sequence Generation on Protein Safety Evaluation Harmful Proteins](<https://www.sota2.com/research/sota/protein-sequence-generation-on-protein-safety-evaluation-harmful-proteins.md>): Protein Sequence Generation; Protein Safety Evaluation Harmful Proteins; BLAST Score: 0.138.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2507.10923>)

[PDF](<https://arxiv.org/pdf/2507.10923.pdf>)

[Implementation](<https://github.com/HICAI-ZJU/KPO>)