Experimental design |
Define intended population |
Patient selection |
|
|
Separate into training and independent validation cohorts |
|
|
Define the anatomical location of skin sampling |
Sample preparation |
Pre-sampling |
Avoid personal care products at the location of sampling for a minimum of 24 h |
|
|
Consider patient medication and use of xenobiotics |
|
Skin VOC collection and storage |
Direct contact vs. non-contact methods |
|
|
Choice of sampling material (cotton/PDMS/SPME) |
|
|
Sampling time |
|
|
Sources of contamination |
|
|
Sample storage (direct analysis or storage at -80 freezer) |
|
Sampling technique |
Consider the use of pre-concentration headspace step or direct injection |
|
|
Use of internal standards in patch |
|
|
QCs (spiking with internal standards to monitor and correct analytical variability) |
MS analysis |
Analytical platform |
Online or offline platforms |
|
|
Optimisation of MS parameters |
|
MS data collection |
Run order (randomisation) and batch effects |
Data Analysis |
Data pre-processing |
Data deconvolution, scaling, peak alignment |
|
|
Removal of irreproducible, contaminant compounds |
|
Statistical analysis |
Descriptive statistics |
|
|
Univariate analysis |
|
|
Multivariate analysis (e.g., PCA, PLS-DA, OPLS-DA) |
|
|
Prediction model (e.g., ROC analysis) |
|
Biological interpretation |
Metabolic pathways (KEGG, HMDB) |