Genome-scale metabolic models (GEMs) are computational representations that enable mathematical exploration of metabolic behaviors within cellular and environmental constraints. Despite their wide usage in biotechnology, biomedicine and fundamental studies, there are many phenotypes that GEMs are unable to correctly predict. GECKO is a method to improve the predictive power of a GEM by incorporating enzymatic constraints using kinetic and omics data. GECKO has enabled reconstruction of enzyme-constrained metabolic models (ecModels) for diverse organisms, which show better predictive performance than conventional GEMs. In this protocol, we describe how to use the latest version GECKO 3.0; the procedure has five stages: (1) expansion from a starting metabolic model to an ecModel structure, (2) integration of enzyme turnover numbers into the ecModel structure, (3) model tuning, (4) integration of proteomics data into the ecModel and (5) simulation and analysis of ecModels. GECKO 3.0 incorporates deep learning-predicted enzyme kinetics, paving the way for improved metabolic models for virtually any organism and cell line in the absence of experimental data. The time of running the whole protocol is organism dependent, e.g., ~5 h for yeast.

基因组规模代谢模型（GEM）是一种计算表示形式，可以在细胞和环境限制下对代谢行为进行数学探索。尽管 GEM 在生物技术、生物医学和基础研究中得到广泛应用，但仍有许多表型是 GEM 无法正确预测的。GECKO 是一种通过使用动力学和组学数据结合酶约束来提高 GEM 预测能力的方法。GECKO 能够重建多种生物体的酶约束代谢模型 (ecModels)，其预测性能比传统的 GEM 更好。在本协议中，我们描述了如何使用最新版本的GECKO 3.0；该过程有五个阶段：(1) 从起始代谢模型扩展到 ecModel 结构，(2) 将酶周转数整合到 ecModel 结构中，(3) 模型调整，(4) 将蛋白质组数据整合到 ecModel 中，以及(5)ecModels的仿真与分析。GECKO 3.0 结合了深度学习预测的酶动力学，为在缺乏实验数据的情况下改进几乎任何生物体和细胞系的代谢模型铺平了道路。运行整个协议的时间取决于生物体，例如酵母约为 5 小时。