Cell-Free Protein Synthesis as a Method to Rapidly Screen Machine Learning-Generated Protease Variants

Aug 5, 2026

Upload by Aurelia

Machine learning tools transform protein prediction and design, yet performance depends on training data quality. Obtaining detailed structural or functional measurements typically requires purification, creating major time and resource burdens at the variant scale needed for ML training. This study introduces cell-free protein synthesis as a fast workflow for screening and scoring protease variants. Screening 48 random variants for initial sampling and 32 targeted variants enabled identification of protease variants with improved kinetic properties.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Abstract

Introduction

Data quality for ML in protein science

Protein fitness landscape and sampling strategies

Need for higher-throughput variant characterization

Edinburgh Research Explorer Cell-Free Protein Synthesis as a Method to Rapidly Screen Machine Learning-Generated Protease Variants Citation for published version: Thornton, EL, Boyle, JT, Laohakunakorn, N & Regan, L 2025, 'Cell-Free Protein Synthesis as a Method to Rapidly Screen Machine Learning-Generated Protease Variants', ACS Synthetic Biology, vol. 14, no. 5, pp. 1710-1718. [https://doi.org/10.1021/acssynbio.5c00062](https://doi.org/10.1021/acssynbio.5c00062) Digital Object Identifier (DOI): 10.1021/acssynbio.5c00062 Link: Link to publication record in Edinburgh Research Explorer Document Version: Publisher's PDF, also known as Version of record Published In: ACS Synthetic Biology General rights Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights. Take down policy The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer content complies with UK legislation. If you believe that the public display of this file breaches copyright please [contact openaccess@ed.ac.uk](contact openaccess@ed.ac.uk) providing details, and we will remove access to the work immediately and investigate your claim. Download date: 08. Jan. 2026 This article is licensed under CC-BY 4.0 [pubs.acs.org/synthbio](pubs.acs.org/synthbio) Research Article Cell-Free Protein Synthesis as a Method to Rapidly Screen Machine Learning-Generated Protease Variants Ella Lucille Thornton, * Jeremy T. Boyle, Nadanai Laohakunakorn, and Lynne Regan * Cite This: [https://doi.org/10.1021/acssynbio.5c00062](https://doi.org/10.1021/acssynbio.5c00062) Read Online ACCESS Metrics & More Article Recommendations *sı Supporting Information ABSTRACT: Machine learning (ML) tools have revolutionized protein structure prediction, engineering, and design, but the best ML tool is only as good as the training data it learns from. To obtain highquality structural or functional data, protein purification is typically required, which is both time and resource consuming, especially at the scale required to train ML tools. Here, we showcase cell-free protein synthesis as a straightforward and fast tool for screening and scoring the activity of protein variants in ML workflows. We demonstrate the utility of the system by improving the kinetic qualities of a protease. By rapidly screening just 48 random variants to initially sample the fitness landscape, followed by 32 more targeted variants, we identified several protease variants with improved kinetic properties. KEYWORDS: cell-free protein synthesis, protein design, machine learning, protease, enzyme activity screen ■ INTRODUCTION The success of machine learning (ML) methods depends on the quality of the data on which an algorithm is trained. The vast amount of high-quality structural data in the protein database (PDB) underlies the remarkable success of the ML program AlphaFold in predicting protein structure from sequence.1 To successfully apply ML methods to other areas of protein science, such as the design of function, high-quality data is vital.2,3 Protein design space can be imagined as a landscape, populated with many different protein sequences each associated with a different “fitness”. The user-defined fitness could be any desirable characteristic such as binding affinity, catalysis rate, yield, or solubility (Figure 1).4 Identifying the protein with the highest fitness within this landscape can be facilitated by using a ML approach to sample and learn from high-quality data from different sequences.5 Employing statistical methods embedded in ML workflows helps the user to avoid common “traps” in the navigation of a protein fitness Figure 1. Diagram representing a hypothetical protein

FAQs

Why does machine learning for protein engineering require high-quality training data?

ML performance depends on the quality of the data used for training, and high-quality functional or structural measurements are vital for learning meaningful patterns.

What problem does purification create in ML-driven protein mutagenesis workflows?

Protein purification is time- and resource-consuming, making it difficult to generate the large variant datasets required for effectively training ML models.

How does the proposed cell-free system improve screening of protease variants?

Cell-free protein synthesis enables rapid screening and scoring of protease variant activity within ML workflows, allowing efficient sampling of the fitness landscape and identification of variants with improved kinetics.

Aug 5, 2026

Upload by Aurelia