What is the reasonable FDR value? 0.05?

[mainharryHR]

Dear Team.

Thank you for the great package. I am curious what is the reasonable FDR value? In the tutorial, one used the FDR = 0.4. In order to publish paper, what is the good cutoff? sccoda_model.set_fdr(sccoda_data_NoCRC, modality_key="No_CRC", est_fdr=0.2)

Thank you all Harry

[johannesostner]

Hi @mainharryHR, this depends on how much false positive risk you want to take. The FDR denotes the expected percentage of your discovered effects that are false positives. Generally, I wouldn't go above 0.2 for a publication. The 0.4 in the tutorial was just for instructive purposes