LIMMA calculates identical adj.P.Val for different proteins from proteomics

Question

[Martin]

Hi, I am trying to calculate statistics for my proteomic data using LIMMA package so I can create some volcano plots. I have normalized log2 transformed intensities with imputed NA values for 2 samples each with 3 biological replicates so 6 columns (+1 annotations). When using LIMMA for some reason several proteins have identical adj.P.Val even tho P.Value is different and logFC and AveExpr is correct and P values seem fine to me. It looks terrible when I use adj.P.Val for volcano plots as several dots are in one line with several lines like that. This happens also with different data and there are several groups like this among whole LIMMA calculated sheet:

ID  logFC   P.Value adj.P.Val
GAA -0,002  0,993   0,997
TUBB2B  -0,005  0,988   0,993
KIF5B   -0,002  0,984   0,993
BPIFB1  0,013   0,986   0,993
ARPC1B  -0,003  0,984   0,993
TXNDC5  0,003   0,985   0,993
KLC4    -0,001  0,989   0,993
...

adj.P.Val are identical even if all decimal places are used...here I just used fewer of them to make it clear

This is my code:

#data loaded to my_data as.matrix with specified rownames and colnames
design <- model.matrix(~gl(2,3))
Limma <-topTable(eBayes(lmFit(my_data, design)), sort.by = "none", number = 1000,adjust.method="BH")

Any idea what is wrong? As I suppose it should be very unlikely to have identical adj.P.Val for different proteins with different values.

Answer 1

[ATpoint]

Please google what the Benjamini-Hochberg correction is and what it does. In a nutshell, it corrects for multiple testing, outputting these adjusted p-values. What you see is normal and expected.

For starters: