Abstract
Metabolomic time course analyses of biofluids are highly relevant for clinical diagnostics. However, some sampling methods suffer from unknown sample sizes commonly known as size effects, which prevents absolute quantification of biomarkers. Recently, studies have developed mathematical post acquisition normalization methods to overcome these problems either by exploiting already known pharmacokinetic information or with statistics.
Here we present an improved normalization method, MIX, that unifies the advantages of both approaches. It combines two normalization terms, one representing pharmacokinetic normalization (PKM) and one representing a popular statistical approach (PQN).
To test the performance of MIX, we generated synthetic data closely resembling real finger sweat metabolome measurements. MIX was subsequently used for normalization of the synthetic data and we were able to show that it overcomes weaknesses of the two normalization strategies applied separately. Moreover, we validate our results by using real finger sweat metabolome data from literature. There, we were able to demonstrate that MIX is more robust than the normalization strategy originally used for the data set.
In conclusion, the MIX method improves the reliability and robustness of biomarker measurements in finger sweat and other biofluids. Moreover, it has potential to pave the way for quantitative biomarker discovery and hypothesis generation from metabolic time course data.
Competing Interest Statement
The authors have declared no competing interest.
Abbreviations
- Symbol
- Name
- asample
- sampling skin area
- b
- part of modified Bateman function
- C, C
- underlying concentration (vector)
- c0
- kinetic parameter
- d
- kinetic parameter
- F
- modified Bateman function
- PQN correction factor
- i
- time point index
- j
- metabolite index
- k
- kinetic parameter
- ℓ
- metabolites used for kinetic fitting
- ℓ+
- metabolites not used for kinetic fitting
- ℒ
- loss
- L
- loss function
- Lag
- kinetic parameter
- measured mass (vector)
- Mref
- reference mass for PQN
- m/z
- mass over charge ratio
- nmetabolites
- number of metabolites
- ntime points
- number of time points
- p
- p-value
- qsweat
- sweat rate
- QC
- median concentration fold change of two samples
- QM
- median mass fold change of two samples
- QPQN
- normalization quotient calculated by PQN
- R2
- coefficient of determination
- rSD
- relative measure of goodness of normalization
- SD
- absolute measure of goodness of normalization
- t
- time
- Vsweat
- collected sweat volume
- v1, v2, v3
- synthetic data sets
- ϵ
- experimental error vector
- θ
- kinetic parameter vector for fitting
- λ
- loss weighting parameter
- τ
- time to collect one sample