Analysis

Unfortunately, there is no standard way to analyse proteome microarray data. While it is an evolving field, eventually, like in the DNA array field, researchers will come to a consensus. But until then, different groups will likely do things in their own preferred way. The aim of these pages is to collect together the different analytical and statistical methods people are using.

A typical data analysis pathway is to first normalize the raw data by subtracting or dividing by the median of the sample-specific IVTT control spots. The latter is called fold-over control (FOC) and often helps comparing data sets from different experiments where reagent dilutions or scanner settings might be slightly different. Many different graphical outputs can be used to display the subtracted or FOC data, depending on the point to be made. See the above drop-down menu ‘Graphical representations‘ for examples. For statistical analysis, log transformed FOC or subtracted data is usually used. For T tests, a non parametric test (eg Wilcoxan rank sum) is used; to correct p-values for false discovery Benjamini-Hochberg or Bonferroni correction can be used. The base stats package in R will do 2-sided T-tests and adjust p-values for false discovery, or you can use other stats packages. Exel is good for graphical output but not really accepted for statistical analysis.