Leveraging Protein Language Model Embeddings for Catalytic Turnover Prediction of Adenylate Kinase Orthologs in a Low-Data Regime

TL;DR

This study demonstrates that protein language model embeddings, especially transformer-based learnable aggregation, improve enzyme catalytic turnover prediction for Adenylate Kinase orthologs in low-data scenarios, outperforming traditional models.

Contribution

It systematically evaluates PLM embeddings and aggregation methods for enzyme activity prediction, highlighting the effectiveness of transformer-based learnable aggregation and the limited benefit of fine-tuning.

Findings

Learned aggregation methods outperform fixed embeddings.

Transformer-based embeddings yield better $k_{cat}$ predictions.

Fine-tuning does not significantly improve performance.

Abstract

Accurate prediction of enzymatic activity from amino acid sequences could drastically accelerate enzyme engineering for applications such as bioremediation and therapeutics development. In recent years, Protein Language Model (PLM) embeddings have been increasingly leveraged as the input into sequence-to-function models. Here, we use consistently collected catalytic turnover observations for 175 orthologs of the enzyme Adenylate Kinase (ADK) as a test case to assess the use of PLMs and their embeddings in enzyme kinetic prediction tasks. In this study, we show that nonlinear probing of PLM embeddings outperforms baseline embeddings (one-hot-encoding) and the specialized $k_{cat}$ (catalytic turnover number) prediction models DLKcat and CatPred. We also compared fixed and learnable aggregation of PLM embeddings for $k_{cat}$ prediction and found that transformer-based learnable aggregation of amino-acid PLM embeddings is generally the most performant. Additionally, we found that ESMC 600M embeddings marginally outperform other PLM embeddings for $k_{cat}$ prediction. We explored Low-Rank Adaptation (LoRA) masked language model fine-tuning and direct fine-tuning for sequence-to-$k_{cat}$ mapping, where we found no difference or a drop in performance compared to zero-shot embeddings, respectively. And we investigated the distinct hidden representations in PLM encoders and found that earlier layer embeddings perform comparable to or worse than the final layer. Overall, this study assesses the state of the field for leveraging PLMs for sequence-to-$k_{cat}$ prediction on a set of diverse ADK orthologs.