VPhaser2 FDR Question

[Richard Orton]

Hi,

We have run VPhaser2 on an Illumina high-depth data set to look at minority variants.

We observe 125 variants in the nofdr.var.txt file but then 129 variants in the fdr.var.txt file.

Just wondering why we observe 4 more variants in the False Discovery Rate file - I would of thought the number would go down due to the FDR correction.

The 4 extra sites in question are:

# Ref_Pos Var Cons Strd_bias_pval Type Var_perc SNP_or_LP_Profile

7836 T C 0.04806 snp 0.3385 A:0:1 C:1684:1553 G:0:1 T:2:9

10088 A T 0.03877 snp 1.151 A:7:30 C:0:1 T:2058:1118

10637 C T 0.04424 snp 0.8096 C:19:8 T:1685:1623

14190 G A 0.04073 snp 0.815 A:1528:904 C:0:1 G:3:17 T:1:0

Checking them out, they have the lowest strd_bias_pval out of all the variants.

Any ideas what i happening (apologies if this is a silly question) - why are these variants not in the noFDR file but are in the FDR file? Is it simply that this is how the Benjamini - Hochberg procedure works, sometimes variants are rejected but sometimes extra variants are passed through.

Cheers,

Richard