Question

DESeq pvalue & padj

0

Entering edit mode

manoharankumar01 • 0

@manoharankumar01-9971

Last seen 7.9 years ago

Hi,

How does pvalue and padj value in DESeq is calculated for samples With no replicates?

gene	baseMeanA	baseMeanB	foldChange	log2FoldChange	pval	padj
Adora2b	151.4009921842	1304.7338537888	8.6177364822	3.1073089836	0.0002	0.0602
Gstt1	32.0034617625	277.8449427122	8.6817152711	3.1179801082	0.0139	0.8028
Fcrl1	12.3090237548	107.2383989415	8.71217743	3.1230333359	0.1017	1

For the genes with same similar fold change pvalue and padj value are high why is that?

Thanks,

DESeq • 2.1k views

ADD COMMENT • link updated 7.9 years ago by Gavin Kelly ▴ 690 • written 7.9 years ago by manoharankumar01 • 0

score 0 · Answer 1 · 2017-02-22

It's in the vignette, and in the help page for DESeq - all samples are treated as replicates (i.e. pooling conditions A and B, and any other conditions you have) for the purposes of estimating the dispersions. This is still done on a gene-by-gene basis, and as dispersion can vary per gene, so can the p-values. The fold-changes are obviously estimated using the original design, it's just the dispersions calculated thus.