Deep learning algorithm aims to accelerate protein engineering

Proteins are the molecular machinery of all living cells and have been exploited for use in many applications, including therapeutics and industrial catalysts. To overcome the limitations of natural proteins, protein engineering is used to improve protein characteristics such as stability and functionality. In a new study, researchers demonstrate a machine learning algorithm that speeds up the process of engineering proteins. The study is published in the journal Nature Communication.

Machine learning algorithms aid in protein engineering by reducing the experimental burden of methods such as directed evolution, which involves multiple rounds of mutagenesis and high throughput screening. They work by simulating and predicting the fitness of all possible target protein sequences after being trained on protein sequence databases.

Although there are many machine learning algorithms, few of them incorporate the evolutionary history of the target protein. This is where ECNet (Context-Integrated Scalable Neural Network), a deep learning algorithm, comes in.

“With ECNet, we are able to examine the target protein and all of its counterparts to see which residues are coupled together and are therefore important for that particular protein,” said Steven L. Miller Chair Professor of Chemical and Biomolecular Engineering Huimin Zhao (BSD leader / CABBI / CGD / GSE / MMG), also director of the Molecule Maker Lab Institute, funded by the National Science Foundation (NSF). “We then combine this information and use the deep learning framework to determine which types of mutations are important for the function of the target protein.”

In a landmark study, researchers showed that ECNet outperforms current methods on several deep mutagenesis datasets. As a follow-up, ECNet was used to design TEM-lactamase TEM-1 – an enzyme that confers resistance to β-lactam antibiotics – and identify variants that had improved fitness and, therefore, were more resistant to ampicillin.

Additionally, ECNet prioritized higher order mutants and new mutants in the analysis. Having a computer tool that can successfully predict higher-order interactions can reduce experimental efforts, Zhao said.

“We combine all the proteins in the database with the specific evolutionary history of the target protein to improve the efficiency of the prediction,” Zhao said. “We can then use the mutants that we generate from our experiences to improve and train the model further. This algorithm is still under development, but it is an overall improvement over what is already known. in the litterature.”

Zhao said researchers are currently using ECNet to develop enzyme catalysts with improved selectivities.

This study was a joint effort with computer science professor Jian Peng (CABBI). Other authors of the study include Yunan Luo, Guangde Jiang, Tianhao Yu, Yang Liu, Lam Vo, Hantian Ding, Yufeng Su, and Wesley Wei Qian.