iCHO2441 genome-scale metabolic model

Creation of updated CHO Genome-scale metabolic model, iCHO2441

iCHO1766 was obtained from the Bigg models database (King et al., 2016), while iCHO2291 was obtained via the BioModels database (Malik-Sheriff et al., 2020). The expanded iCHO2441 GeM was constructed by coupling the secretory machinery presented within the iCHO2048 (Gutierrez et al., 2020a) to the recently published updated iCHO2291 (Yeo et al., 2020a). This was achieved by adapting the Jupyter Notebooks developed by Gutierrez et al. (2020) to use the updated iCHO2291 as a base model to which secretory reactions may be added. In brief, information of each secreted product including amino acid composition, presence of a signal peptide, number of disulphide bonds, number of core N-linked glycans, and molecular weight were fed to the notebook and used to add the appropriate secretory pathway reactions to the model (Table S1). For intracellular flux prediction, a custom model was generated for each secreted product using product composition data. For auxotrophy and gene essentiality predictions, a generic IgG structure was used to add secretory reactions to the model (Table S1). As per the recommendation by Hart et al., genes with transformed values less than -3 were classed as ‘not expressed/ negative confidence’ (class -1). Genes with transformed values between -3 and -1.5 were classed as ‘low confidence expressed’ (class 1). Genes with transformed values between -1.5 and 0 were classed as ‘medium confidence expressed’ (class 2). Genes with transformed values greater than 0 were classed as ‘expressed/high confidence’ (class 3). These classifications were mapped to reactions within iCHO2441 using the method described above, where unclassified reactions were given a class of 0. Reconstructions were generated per experiment using the CORDA function for python which took iCHO2441 constrained with experimental uptakes (Table S2) as a base model and the gene confidence classifications as inputs. An in-depth rationale behind this novel approach is presented in supplementary material 1.

构建更新版的CHO基因组级代谢模型iCHO2441
iCHO1766源自Bigg模型数据库（King等，2016年），而iCHO2291则通过BioModels数据库获取（Malik-Sheriff等，2020年）。扩展的iCHO2441基因组级代谢模型通过将iCHO2048（Gutierrez等，2020a）中呈现的分泌机制与近期发表的更新版iCHO2291（Yeo等，2020a）相耦合而构建。此过程通过修改Gutierrez等（2020）开发的Jupyter笔记本实现，以更新版的iCHO2291作为基础模型，并向其中添加分泌反应。概括而言，每个分泌产物的信息，包括氨基酸组成、信号肽的存在、二硫键的数量、核心N-连接甘露糖胺的数量和分子量，均输入至笔记本中，用以向模型添加相应的分泌途径反应（见表S1）。为预测细胞内流量，针对每个分泌产物生成了一个定制的模型，并使用产物组成数据进行。为预测营养缺陷性和基因必要性，采用通用的IgG结构向模型添加分泌反应（见表S1）。依据Hart等人的建议，转换值小于-3的基因被归类为“未表达/低置信度”（类别-1）。转换值介于-3和-1.5之间的基因被归类为“低置信度表达”（类别1）。转换值介于-1.5和0之间的基因被归类为“中等置信度表达”（类别2）。转换值大于0的基因被归类为“表达/高置信度”（类别3）。这些分类通过上述方法映射至iCHO2441中的反应，其中未分类的反应被赋予类别0。通过使用Corda函数为Python生成每个实验的重建模型，该函数以受实验摄取约束的iCHO2441作为基础模型，并将基因置信度分类作为输入。这种创新方法的深入阐释见补充材料1。