suprisingly high logFC using limma

Question

[alwayshope]

Dear guys,

Do you have any experience when using limma for one batch of proteomics data, the log2FC is surprisingly high (1500-3000), but when checking the expression matrix, the value among the samples are pretty similar.

The design is ~0+ treatment, which seems quite normal, any possible reasons? It does work well for other data, quite strange...

Many thanks!

[ATpoint]

You will need to show data, code and some plots to illustrate.

Answer 1

[Gordon Smyth]

It's a sure sign the data has been input to limma incorrectly.

[alwayshope]

Thanks a lot! Turns out the input matrix has not been normalized...

Thank you all very much!