Computing the FDR

The program NManofdr2 implements the procedure of Benjamini and Hochberg (1) to determine the p-value cutoffs corresponding to low, moderate and large FDR values. Let's run the program

$NManofdr2 -i S5/NManova -d HucH3K14_norm.db -o S5/FDR.txt

The output of the program is stored in the file FDR.txt which is a simple table like this:

$cat S5/FDR.txt
###
# Output of NManofdr2
###
FDR	Pval	#SNEP	#NUC	%SNEP
1.00e-04	2.89e-06	1822	62824	0.02900
1.00e-03	4.74e-05	2984	62824	0.04750
1.00e-02	8.44e-04	5303	62824	0.08441
5.00e-02	6.84e-03	8602	62824	0.13692
1.00e-01	1.75e-02	10997	62824	0.17504

where the first column provides the FDR value, the second one the corresponding p-value cut-off, the third one indicates the corresponding number of SNEPs (i.e. nucleosomes for which the p-value from the ANOVA is lower than the p-value cut-off), the fourth has the same value for all rows and corresponds to the total number of nucleosome having a valid p-value and the last column is just the fraction of SNEPS, i.e the ratio between the third and the fourth column.