Fig. 3

Quartet-based signal-to-noise ratio enables intra-laboratory reliability assessment. a, b Reliability assessment using signal-to-noise ratio (SNR) in different laboratories with (a) or without (b) filtration. The results were visualized by PCA plots. The number of features used and the calculated SNR were shown above each plot. c Schematic diagram of SNR calculated as the ratio of the averaged distance between different Quartet samples (“signal”) to the averaged distances between technical replicates for each sample (“noise”) on a 2D-PCA scatter plot. d SNRs calculated with metabolites filtered with different criteria across all laboratories. e The inter-laboratory data quality was ranked using different QC metrics, including the percentages of retained metabolites using different filtering criteria (CV < 30%; ICC > 0.4; CV < 30% and ICC > 0.4), PCC of technical replicates, and SNR calculated with or without filtering. f The concordances (Spearman Rho) of data quality ranking using different QC metrics