Skip to main content

QMDB data interpretation

Are these concentration ranges really relevant/comparable to our own measurement results?​

If you measured your human EDTA-plasma samples with a biocrates AbsoluteIDQ p180 or MxP Quant500 kit and target normalized the concentrations to the biocrates quality control level 2 (according to the kit protocol), there is a good chance that the concentration ranges provided by the QMDB are comparable to your own measurement results. You may need to set the available filter options to make the donor population more comparable to the demographics of your own study population. This has been addressed in our Application note on the QMDB (coming soon).

Can I omit a healthy control group in my study and use the QMDB reference ranges as a control instead?​

We have shown that the concentration ranges of the QMDB are comparable to external studies and can be used as control group when no internal controls are available. For an optimal study design, however, it is still recommended to use an internal control group if available.

According to standard literature, a reference sample group is an adequate number of selected persons to represent a well-defined reference population. Does this apply for the QMDB?​

The QMDB strives to provide relevant metabolite concentration ranges that can be considered normal in the healthy population. As the demographics of the QMDB donors are not representative of the demographics in the general healthy population, the mean and median concentrations are likely to somewhat deviate from the true mean and median in the healthy population.

Have you tested the suitability of the QMDB data as reference ranges?​

Yes. We have compared the reference ranges to several independent datasets, representing subsets of the healthy population, and found congruence in reasonable limits. You find details on this in our Application note on the QMDB (coming soon).

A reference interval usually comprises a fraction of the values measured in reference individuals, most frequently the central 95% of the distribution located between the 0.025 and 0.975 percentiles. Does this apply for the QMDB?​

We decided that the QMDB should include all values measured in the samples and not to cut off the highest and lowest 2.5% of the values. The provided descriptive statistics values (mean, median, 1st and 3rd quartile) provide additional information on the value dispersion within the full value range.

When using the statistical comparison in the Excel template comparing my diseased study group to the QMDB contents, why do I get so many significant differences?​

The QMDB contains a lot of samples, and with a high number of values, even small differences can become statistically significant. When working with comparisons including QMDB exports with hundreds of samples, you should not use p<0.05 to determine significance. We have made good experiences with setting a minimum fold change, e.g. of 1.5, as a significance criterium together with a q value (p value corrected for multiple testing) of 0.001.

I compared the concentration ranges of my control group to the QMDB ranges. Why are there so many significant differences and what can I do?​

Usually, the QMDB values should match with a healthy human control group quite well. We have assessed this in our Application note (coming soon). Have you filtered the QMDB so that the demographics correspond to your study? Have you adjusted the significance criteria to include a minimum fold change and a reasonable q value, as elaborated elsewhere in this FAQ section? If you did this and still see your control group is very different from the QMDB dataset, it is often for one or more of the following reasons:

  • Your data was not normalized in the most suitable way. We recommend target normalizing your data to the biocrates QC level 2 for maximum comparability
  • Your control group has a markedly higher fraction of lipemic samples. We have observed this sometimes when looking at other control groups and there is some fluctuation. Try excluding samples with markedly higher lipid levels than usual from your dataset
  • Your sample quality is below average. If too much time passed between blood drawing and centrifugation to obtain plasma, or between blood drawing and sample freezing, metabolic changes occur in the samples affecting the concentrations of many metabolites. This also occurs if samples underwent repeated freeze-thaw cycles. If you have details on preanalytics, exclude all samples with >2 h passed before centrifugation and >6 h passed before freezing, and with more than 3 freeze-thaw cycles

Is it allowed to publish results comprising exported data from the QMDB?​

Yes, you may publish data exported from the QMDB as long as you hold a valid QMDB license, and you cite the QMDB correctly like this: “Data Source: Quantitative Metabolomics Database (QMDB) on ‘Date’; https://biocrates.com/quantitative-metabolomics-database". Please also cite our Application note (coming soon).