《生命科学》 2026, 38(2): 357-366
人工智能驱动的生物制造进展与趋势
摘 要：
生物制造正经历从经验驱动向数据驱动的范式转变，人工智能在其中发挥着重要作用。本文综述了2025年人工智能在生物制造核心环节——包括生物元件设计、代谢网络建模、人工生命系统与无细胞合成系统构建、工艺优化与过程控制等方面的最新进展。尽管当前仍面临生物机制认知局限、数据稀缺、模型可解释性不足等挑战，未来随着人工智能与合成生物学、自动化技术的深度融合，将推动生物制造向可编程、智能化、可持续的新一代工业体系跨越，为医药、化工等领域的绿色转型提供核心驱动力。
通讯作者：毛开云 , Email:kymao@sinh.ac.cn
Abstract:
Biomanufacturing is undergoing a profound paradigm shift from traditional experience-driven methods to innovative data-driven approaches, with artificial intelligence (AI) serving as a central enabling force. This article systematically reviews breakthroughs and applications of AI across key domains of biomanufacturing. In biological component design, generative AI facilitates “from-scratch” innovation for creating DNA regulatory elements, signal peptides, and enzymes, while continuously refining functionality through intelligent “design-build-test-learn” closed-loop systems. This shift enables researchers to move beyond imitation of natural systems toward truly rational and programmable biological design. In metabolic network reconstruction, multi-scale metabolic models combined with advanced deep learning frameworks like AlphaGEM enable dynamic simulation of complex metabolic behaviors, cross-species generalization, and exploration of previously uncharted biological functions. These computational advancements are particularly significant for engineering non-model organisms with industrial relevance, where traditional characterization methods are often inadequate. Within cell-free synthesis systems, AI converges with microfluidic and automation platforms to establish high-throughput, intelligent, and self-optimizing protein production pipelines, effectively decoupling biological synthesis from the constraints of cellular viability and regulation. This approach enables rapid prototyping of biomolecules that would be difficult or impossible to produce using conventional cellular systems. For process optimization and control, AI drives the evolution toward rationally designed, dynamically optimized, and autonomously regulated production through predictive modeling, real-time sensor integration, and adaptive control strategies. These systems continuously learn from process data to maximize yield, minimize variability, and ensure consistent product quality while reducing resource consumption and operational costs. While AI has substantially accelerated biomanufacturing research and development (R&D) and enhanced production efficiency, several challenges remain, including incomplete understanding of underlying biological mechanisms, scarcity of high-quality annotated datasets, limited model interpretability (often described as the “black box” problem), complexities in multi-objective optimization, and stringent real-time operational demands in industrial settings. Looking forward, deeper integration of AI with synthetic biology, robotic automation, and multi-omics data analytics will propel biomanufacturing toward a programmable, intelligent, and sustainable next-generation industrial ecosystem. Future advancements are expected to include more sophisticated physics-informed neural networks that incorporate fundamental biological principles, federated learning approaches to leverage distributed data while maintaining privacy, and the development of digital twins for entire bioprocesses. This convergence is poised to deliver a key technological engine for the green transformation of critical sectors such as pharmaceuticals, chemicals, and energy. Ultimately, the synergy between AI and biotechnology promises to establish a more efficient, resilient, and environmentally conscious bioeconomy capable of addressing global challenges in health, materials, and sustainable production.
Communication Author：MAO Kai-Yun , Email:kymao@sinh.ac.cn
SciCloud投稿系统