Recent advances in metabolite quantification and identification have enabled new research into the detection and control of titer inhibitors and promoters. This paper presents a bigdata analytics study to identify both inhibitors and promoters using multivariate data analysis of metabolomics data. By applying multi-way partial least squares (PLS) model to metabolite data from four fed-batch bioreactor conditions where feed formulation and selection agent concentrations varied, metabolites which exhibited the most significant impact on titer during cultivation were ranked from highest to lowest. The model outputs were then constrained to reduce the number of statistically relevant inhibitors or promoters to the top ten, which were used to conduct metabolic pathway analysis. Furthermore, a method is presented for identifying amino acids that prevent the accumulation of the inhibitors and/or enhance the formation of promoters during production. Finally, the metabolomics and pathway analysis results were integrated and validated with transcriptomics data to characterize metabolic changes occurring among different growth conditions. From these results, new feeding strategies were implemented which resulted in increased fed-batch production titer. Methodology from this work could be applied to future process optimization strategies for biotherapeutic production.

中文翻译：

---

Bigdata Analytics识别代谢抑制剂和促进剂，以提高生产率并优化单克隆抗体（mAb）的生产过程

代谢物定量和鉴定的最新进展使新的研究能够用于滴度抑制剂和启动子的检测和控制。本文提出了一项大数据分析研究，以利用代谢组学数据的多变量数据分析来识别抑制剂和启动子。通过将多元偏最小二乘（PLS）模型应用于四种补料分批生物反应器条件下的代谢物数据，其中饲料配方和选择剂的浓度会发生变化，对培养期间效价影响最大的代谢物从最高到最低排名。然后限制模型的输出，以将统计学上相关的抑制剂或启动子的数量减少到前十名，将其用于进行代谢途径分析。此外，提出了鉴定在生产过程中防止抑制剂积累和/或增强启动子形成的氨基酸的方法。最后，将代谢组学和途径分析结果进行整合，并用转录组学数据进行验证，以表征不同生长条件之间发生的代谢变化。根据这些结果，实施了新的饲喂策略，从而提高了批量生产的滴度。这项工作的方法论可以应用于生物治疗生产的未来工艺优化策略。整合了代谢组学和通路分析结果，并用转录组学数据进行了验证，以表征不同生长条件下发生的代谢变化。根据这些结果，实施了新的饲喂策略，从而提高了批量生产的滴度。这项工作的方法论可以应用于生物治疗生产的未来工艺优化策略。整合了代谢组学和通路分析结果，并用转录组学数据进行了验证，以表征不同生长条件下发生的代谢变化。根据这些结果，实施了新的饲喂策略，从而提高了批量生产的滴度。这项工作的方法论可以应用于生物治疗生产的未来工艺优化策略。