Improving Protein Sequence Design through Designability Preference Optimization
This work addresses the challenge of ensuring designability in protein sequence design for computational biology, representing an incremental improvement over existing methods.
The paper tackles the problem of protein sequence design by shifting the training objective from sequence recovery to designability, using preference optimization to improve the likelihood that designed sequences fold correctly. The result is a nearly 3-fold increase in in silico design success rate, from 6.56% to 17.57% on a challenging benchmark.
Protein sequence design methods have demonstrated strong performance in sequence generation for de novo protein design. However, as the training objective was sequence recovery, it does not guarantee designability--the likelihood that a designed sequence folds into the desired structure. To bridge this gap, we redefine the training objective by steering sequence generation toward high designability. To do this, we integrate Direct Preference Optimization (DPO), using AlphaFold pLDDT scores as the preference signal, which significantly improves the in silico design success rate. To further refine sequence generation at a finer, residue-level granularity, we introduce Residue-level Designability Preference Optimization (ResiDPO), which applies residue-level structural rewards and decouples optimization across residues. This enables direct improvement in designability while preserving regions that already perform well. Using a curated dataset with residue-level annotations, we fine-tune LigandMPNN with ResiDPO to obtain EnhancedMPNN, which achieves a nearly 3-fold increase in in silico design success rate (from 6.56% to 17.57%) on a challenging enzyme design benchmark.