Hi there,

I'm trying to calculate FDR-corrected p-values from the p field in my ROI.mat file. The p field in this file contains p-values which are equivalent on each side of the diagonal, as expected. However when I compute the FDR-corrected values for some reason the corrected p-values differ on each side of the diagonal (I've attached a screenshot).  

I'm using p_FDR=conn_fdr(p_uncorrected) to compute the FDR-corrected p-values.

If anybody has any thoughts on the values might be different on each side of the diagonal I would appreciate it!

Alex

Hi Alex,

When entering a matrix of uncorrected p-values conn_fdr will compute p-FDR values separately for each column (values along the first-dimension) of your matrix. So, in this case, that means that p_FDR, as you are computing it now, will compute seed-level FDR-corrected p-values (the multiple-comparison correction is correcting for multiple target ROIs, but separately for each seed ROI), which is not a symmetric measure, instead of analysis-level FDR-corrected p-values (the multiple-comparison correction is correcting for all pairs of seed&target ROIs), which is a symmetric measure. If you want to compute analysis-level p-fdr values, simply use the syntax:

p_FDR = reshape(conn_fdr(p_uncorrected(:)),size(p_uncorrected));

Hope this helps
Alfonso
Originally posted by Alexandra Cross: