AI-powered high-throughput digital colony picker platform for sorting microbial strains by multi-modal phenotypes

作者：Zhidian Diao, Qiqun Peng, Sijun Luo, Lingyan Kan, Anle Ge, Wei Gao, Runxia Li, Weiwei Bao, Xixian Wang, Yuetong Ji, Jian Xu, Shihui Yang, Bo Ma · 发表于：Nature Communications · 年份：2025 · DOI：10.1038/s41467-025-63929-7 · 被引用次数：17 · 研究领域：Innovative Microfluidic and Catalytic Techniques Innovation、Cell Image Analysis Techniques、3D Printing in Biomedical Research

Phenotype-based screening remains a major bottleneck in the development of microbial cell factories. Here, we present a Digital Colony Picker (DCP), an AI-powered platform for automated, high-throughput screening and export of microbial clones based on growth and metabolic phenotypes at single-cell resolution, without agar or physical contact. Using a microfluidic chip comprising 16,000 addressable picoliter-scale microchambers, individual cells are compartmentalized, dynamically monitored by AI-driven image analysis, and selectively exported via laser-induced bubble technique. Applied to Zymomonas mobilis, DCP enabled en masse screening and identified a mutant with 19.7% increased lactate production and 77.0% enhanced growth under 30 g/L lactate stress. This phenotype was linked to overexpression of ZMOp39x027, a canonical outer membrane autotransporter that promotes lactate transport and cell proliferation under stress. DCP provides a multi-modal phenotyping solution with spatiotemporal precision and scalable throughput, offering a generalizable strategy for accelerated strain engineering and functional gene discovery. Phenotype-based screening is a major bottleneck in the development of microbial cell factories. Here the authors build an AI-powered digital colony picker for single-cell-resolved, contactless screening and export of microbial strains, which identified lactate-tolerant Zymomonas mobilis mutants.