Variability assessment of LC-MS experiments and its application to experimental design and difference detection

In quantitative liquid chromatography-mass spectrometry (LC-MS) experiments, variability assessment helps improve experimental design and detect true differences in ion abundance. A peak-level mixed effects model is considered to estimate the variability due to heterogeneity of the biological samples, inconsistency in sample preparation, and instrument variation. We focus on determining the optimal number of replicates to achieve adequate statistical power. We perform two simulation studies to demonstrate important factors in replication assignment, sample size calculation and difference detection. The parameters of the simulation studies are derived based on analysis of an in-house LC-MS data set. Sensitivity and false discovery rate of the mixed effects model are compared to those of t-test and fixed effects model.