search

const-ae / proDA

Protein Differential Abundance for Label-Free Mass Spectrometry https://const-ae.github.io/proDA/
17 stars 8 forks source link

different results for some proteins depending test used. #14

Open wolski opened 2 years ago

wolski commented 2 years ago

Hi Constantin,

A Question:

I am running two tests with proDA, once the LR and once the Wald test. The data is a repeated measurement on 5 subjects with two groups (high and low)

I would think that the two tests are equivalent. However, for a group of proteins, I am getting substantially different results. Can you have a look at my code, please? Is the difference in p-values within the range you would expect, and what is the explanation for it in your opinion?

fit <- proDA(se, design = ~ group_  + subject_, data_is_log_transformed = TRUE)
lrTest <- test_diff(fit, reduced_model =  ~ subject_)
fit2 <- proDA(se, design = ~ group_  + subject_ - 1, data_is_log_transformed = TRUE)
waldTest <- test_diff(fit2, group_high - group_low)
lrT <- select(lrTest, name, adj_pval) 
wT <- select(waldTest, name, adj_pval)
tmp <- inner_join(lrT, wT, by = "name", suffix=c("lr.Test","wald.Test"))
plot(tmp$adj_pvallr.Test, tmp$adj_pvalwald.Test)

image

best regards Witold