“…We are aware of other and more sophisticated normalization techniques, and we furthermore understand the necessity to normalize when pooling metabolomics data across studies. 18,19 In addition, we acknowledge that typical metabolomics data processing pipelines include a data filtering step, often using the '80% rule' which removes features that have missing data in more than 20% of the samples, 20 before performing normalization. However, the use of normalization factors obtained after such a data filtering procedure did not seem to affect the observed correlations that much, at least not for this dataset (see Fig.…”