Table IV Comparing methods of significance testing with the 10 versus 30 °C data set

Three methods for estimating differential protein abundance were compared.

Student's t testaUnmoderated t testbModerated t testc
pdBonfeFDRfpBonfFDRpBonfFDR
Ng9549549548308308301,1721,1721,172
<0.05h3255627214414582141145
E(FP)i480.0514420.053590.052
FDR (%)j150.15290.45280.55
  • a Student's t test where each protein abundance measurement was treated as independent using a simple linear model accounting for 14N and 15N label reversal only.

  • b Unmoderated t test with the full linear model.

  • c Empirical Bayes moderated t test with the full linear model.

  • d p, the number of differentially abundant proteins with statistic <0.05 for unadjusted p values.

  • e Bonf, the number of differentially abundant proteins with statistic <0.05 for Bonferroni method-corrected p values.

  • f FDR, the number of differentially abundant proteins with statistic <0.05 for FDR-corrected q values.

  • g The number of proteins analyzed from the 1,172 quantified proteins; proteins analyzed by the Student's t test had at least two observations, the unmoderated t test had at least three observations or two observations from the same extraction buffer, and the moderated t test had at least one observation.

  • h Number of differentially abundant proteins with statistic <0.05.

  • i Expected false positives.

  • j FDR was calculated by dividing E(FP) by the number of differentially abundant proteins with statistic <0.05.