Achieving Effective Batch-to-Batch Error Correction through Suppression Correction and Dual MSTUS Normalization

Batch effect corrections are necessary to remove instrumentally introduced variance but should retain biological variance. Most common batch correction techniques are based on analysis of quality control (QC) samples, sample amplitude normalization, or other statistical and normalization techniques. Many of these approaches correct for the totality of the MS signal and do not differentiate between analytical and biological variances. In this study, Saccharomyces cerevisiae was treated with four different oxidants at five different time points. Biologically replicated sample sets were analyzed across six batches where the batch effect was intentionally introduced. We used the Isotopic Ratio Outlier Analysis (IROA) workflow method to first correct suppressed raw MS data for source related errors and then subsequently applied an enhanced normalization technique. These results were compared to those of other widely used normalization methods. Importantly, for validation of our approach, we conducted careful statistical evaluation and biological validation to ensure that the correction had effectively removed batch effects without losing biological signal. This workflow method could successfully distinguish between technical and biological variation and reduce the relative standard deviation (RSD) error to 1% compared to raw data, a clear indication that this method can be used to reliably enhance data quality. The experimental design afforded us with a means to test the ability of each normalization method to correctly group similar samples in a "data fidelity test" that may have utility for future studies.