From Structure to Function: Preference Alignment for Function-Aware Protein Inverse Folding

Protein inverse folding (IF) models conditioned on structure achieve high sequence recovery but often fail to preserve biological function due to the lack of functional supervision. We propose a Function-aware Preference Alignment (FPA) framework that avoids explicit function optimization by fine-tuning IF models to prefer function-preserving sequences over function-disrupting alternatives. Our approach constructs reliable preference pairs in silico using hypothesis-driven perturbations of critical residues and model-consistent likelihood constraints, enabling scalable supervision without additional wet-lab measurements. Our FPA framework guides protein sequence design models toward generating sequences that better preserve functional integrity, while remaining model-agnostic and compatible with existing inverse folding pipelines such as ProteinMPNN and ESM-IF. Extensive experiments on protein design benchmarks show that our fine-tuned models significantly outperform pretrained counterparts in preserving functional integrity during protein sequence design.