Design and Analysis of Metabolomic Experiments:
A Statistician's Perspective

Dr Jim McNicol

Biomathematics & Statistics Scotland, SCRI, Invergowrie, Dundee, DD2 5DA, UK.

Important points concerning design, replication and randomisation of metabolomic experiments will be discussed. A general two-stage procedure for the analysis of designed experiments will be illustrated. The first stage involves screening the complete data set to identify the major sources of variation, intended and unintended. The second stage involves grouping differentially expressed individual compounds according to their significance profiles across the experimental treatments. A simple graphical procedure to link results from an experiment run on two different metabolic platforms will be presented.