The advent of systems biology has enabled routine generation of large multivariate data sets for the construction of mechanistic models which can be used to gain insights into biological processes. These large data sets, however, are often accompanied by significant background levels, making robust biological interpretation of the results challenging. In this study, we performed a rigorous characterization of biological and technical variance in time course metabolomics data from bioreactor cultivated Chinese hamster ovary (CHO) cells. Samples were collected at five time points from two replicate lab-scale and one pilot-scale bioreactors. We identified a number of metabolites and metabolic pathways which exhibited low biological variance consistently across different time points, whereas none showed consistently high biological variance. Interestingly, biological variance only accounted for about 0.4% of the total variance between the lab-scale duplicate bioreactors while contributions from technical variance were 9-12%. Using these variances as the backdrop, only 20-25 instead of 65-105 of the 267-387 detected metabolites showed significant differences in concentration across bioreactor scales. Furthermore, background removal also showed tighter clustering of samples, both over time and bioreactor scale and no substantial differences in metabolic profiles were observed. In contrast, if these variances were not taken into consideration, the samples were separated to some degree with either time or scale; and aminoacyl t-RNA biosynthesis and nitrogen metabolism were identified as differentially expressed. Correcting for inherent biological variability enabled determination of the true biological differences between bioreactor scales and can improve biological interpretation from metabolomics experiments.

Characterizing technical and biological variance in CHO cell time-series metabolomics data

249th National Meeting (2015)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.