DisProtEdit: Exploring Disentangled Representations for Multi-Attribute Protein Editing

TL;DR

DisProtEdit is a novel framework for controllable protein editing that learns disentangled representations of structure and function using dual-language supervision, enabling interpretable and modular modifications.

Contribution

It introduces a new disentangled representation learning approach for protein editing with a large-scale multimodal dataset and achieves improved interpretability and control.

Findings

Competitive performance on protein editing benchmarks.

Achieves up to 61.7% success rate on multi-attribute editing.

Demonstrates improved interpretability and controllability.

Abstract

We introduce DisProtEdit, a controllable protein editing framework that leverages dual-channel natural language supervision to learn disentangled representations of structural and functional properties. Unlike prior approaches that rely on joint holistic embeddings, DisProtEdit explicitly separates semantic factors, enabling modular and interpretable control. To support this, we construct SwissProtDis, a large-scale multimodal dataset where each protein sequence is paired with two textual descriptions, one for structure and one for function, automatically decomposed using a large language model. DisProtEdit aligns protein and text embeddings using alignment and uniformity objectives, while a disentanglement loss promotes independence between structural and functional semantics. At inference time, protein editing is performed by modifying one or both text inputs and decoding from the updated latent representation. Experiments on protein editing and representation learning benchmarks demonstrate that DisProtEdit performs competitively with existing methods while providing improved interpretability and controllability. On a newly constructed multi-attribute editing benchmark, the model achieves a both-hit success rate of up to 61.7%, highlighting its effectiveness in coordinating simultaneous structural and functional edits.