Design of biomolecular sequences by artificial intelligence

Ye Wang, Haochen Wang, YAN Minghao, Hu Guanhua, Xiaowo Wang

Abstract:

Synthetic biology research adheres to the principles of learning from nature, transforming nature, and transcending nature. Its core is to artificially optimize, modify, and recombine genetic elements to obtain artificial biological systems that meet specific needs. Acquiring biological components with excellent performance is the foundation for constructing and controlling artificial biological systems.
In recent years, artificial biomolecules have been widely used in fields such as metabolic engineering and gene therapy. How to efficiently search and design molecular sequences with specific biological functions in the vast molecular sequence space is a significant scientific problem facing synthetic biology. With the rapid development of artificial intelligence technology, intelligent algorithms have shown great potential in mining complex biological features and designing biomolecules.
This article focuses on analyzing the application characteristics of deep generative models in the design of different artificial biological sequences, from the perspective of using complex feature patterns discovered by deep learning technology as guidance to intelligently explore new drug molecules, nucleic acid sequences, and protein sequence spaces. On this basis, combined with application cases of biomolecule design such as small molecule compounds, nucleic acids, and proteins, it summarizes and analyzes the directional optimization strategies for artificial biomolecule sequence design. In order to evaluate the molecules designed by intelligent algorithms, it systematically analyzes the characteristics of sequence design evaluation schemes from different perspectives in different fields, and prospects the development of intelligent design of artificial biological sequences, which needs to fully consider the complex characteristics of biological systems with multi-level regulation and high coupling, and optimize the design of biological sequences at different levels from a system perspective, so as to promote the intelligent adaptation and optimization of artificial biological systems.