Figure 5
Statistical analysis and decision-making tree used to identify protein expression changes in glycoprotein enriched fractions by a label-free LC-MS approach. (A) Comparison of the number of differentially-expressed multiply-charged ions identified when different cut-off values for fold change (displayed as |log2(TGF-β/CTL)|) and p value are used (gray bars). As an estimate of the number of false positives identified with each set of cutoff values, two separately prepared control samples (CTL-1 and CTL-2) were analyzed in a similar fashion (black bars). The difference between the number of differentially-expressed ions identified in the TGF-β/CTL comparison and the CTL-2/CTL-1 comparison is shown above each set of bars and represents the number of real, truly differentially expressed ions. (B) The decision tree used to determine whether a given protein is differentially-expressed is shown. Briefly, all peptides that were identified in any given biological repeats were used to determine initial statistical significance. Subsequent filtering then required that differentially-expressed peptides were identified in at least 2 biological repeats and that the remaining biological repeat showed a consistent trend towards up- or down-regulation.