pLM-bVAE: protein language model-enhanced binary representation improves quantum annealing-based protein sequence design

pLM-bVAE improves QA-based protein sequence design by producing binary encodings and reconstructed sequences with higher accuracy than bVAE across three public protein variant datasets.

• Why it matters: QA optimization depends on how well protein sequences are converted into binary vectors and then decoded back, but bVAE’s protein suitability was not systematically tested.
• What they did: The authors augment bVAE with binary representations derived from a pretrained protein language model, then compare reconstruction and encoding performance on three public variant datasets and embed pLM-bVAE in a QA design pipeline.
• The result: In all three datasets, pLM-bVAE yields more high-scoring sequences with greater dissimilarity from training and validation data than bVAE, supporting its use for QA-based design.

AI summary of the abstract · may contain errors · not medical advice · read the paper.