The case for well-conducted experiments to validate statistical protocols for 2D gels: different pre-processing = different lists of significant proteins

Abstract Background The proteomics literature has seen a proliferation of publications that seek to apply the rapidly improving technology of 2D gels to study various biological systems. However, there is a dearth of systematic studies that have investigated appropriate statistical approaches to analyse the data from these experiments. Results Comparison of the effects of statistical pre-processing on the results of two sample t-tests suggests that the results of 2D gel experiments and by extension the conclusions derived from these experiments are not independent of the statistical protocol used. Conclusions This study suggests that there is a need for well-conducted validation studies to establish optimal statistical techniques to be used on such data sets.