LevSeq: Rapid Generation of Sequence-Function Data for Directed Evolution and Machine Learning

Area of Science:

• Biochemistry
• Molecular Biology
• Bioinformatics

Background:

• Sequence-function data is crucial for understanding protein landscapes but is seldom generated during directed evolution.
• Existing methods for obtaining comprehensive sequence-function data are often time-consuming and resource-intensive.

Purpose of the Study:

• To introduce Long-read every variant Sequencing (LevSeq), a novel pipeline for rapid generation of protein sequence-function data.
• To enable data-driven protein engineering and machine learning-guided protein engineering (MLPE).

Main Methods:

• LevSeq utilizes a dual barcoding strategy combined with nanopore sequencing.
• The pipeline integrates seamlessly into existing protein engineering workflows.
• Open-source software is provided for data analysis and visualization.

Main Results:

• LevSeq rapidly generates sequence-function data for entire protein-coding genes.
• The method facilitates quality control of mutagenesis libraries, reducing costs and time.
• Simulation studies confirm LevSeq's accuracy in variant detection under diverse conditions.
• LevSeq demonstrated utility in engineering protoglobins for novel chemistry.

Conclusions:

• LevSeq empowers data-driven protein engineering by providing essential sequence-function data.
• Widespread adoption and data sharing will advance the understanding of protein sequence-function relationships.
• The pipeline enhances efficiency and reduces resource expenditure in protein engineering efforts.