Question

Interpreting results of DESeq2, regarding baseMean and log2FoldChange.

1

Entering edit mode

blastzone.heimerdinger ▴ 20

@blastzoneheimerdinger-15116

Last seen 6.3 years ago

Dear forum,

Hello, I have a question regarding interpreting DESeq2 results regarding baseMean and log2FoldChange. I run DESeq() to perform differential expression analysis between group 1 and 2, and got the results like following, condition is c("1", "2").

	baseMean	log2FoldChange	lfcSE
A	1000	-0.80	0.15
B	400	0.60	0.14
C	500	0.80	0.20

I have a difficulty interpreting these, as it seems that baseMean represents mean of all participants (both group 1 and 2). I thought that "baseMean represents mean of group 1 (base level), and log2FoldChange indicate the change of group2 from group1." Is this interpretation correct?

If correct, How should I interpret this baseMean? Is it just showing all the participants mean for reference? I don't think the result is showing that "log2FoldChange of group 2 from all the group baseMean". I hope I clarified my question well...

Thanks in advance.

deseq2 • 6.6k views

ADD COMMENT • link updated 6.7 years ago by Michael Love 43k • written 6.7 years ago by blastzone.heimerdinger ▴ 20

score 1 · Answer 1 · 2018-03-30

1

Entering edit mode

Michael Love 43k

@mikelove

Last seen 4 days ago

United States

See this link below which describes the variables:

https://bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#access-to-all-calculated-values

It’s the mean of all samples in the dataset. We kept the same name as DESeq to avoid breaking MA plot code but there isn’t always a reference level in DESeq2 so we just use the mean of all normalized counts.

ADD COMMENT • link 6.7 years ago Michael Love 43k

0

Entering edit mode

Thank you very much for the answer and pointing me to the right vignette! I read the section and still get confused about log2FoldChange ... if I understand it correctly, ignoring the baseMean of all the samples displayed in the results() table, from the results we can conclude that "if baseMean of A in group 1 is 100, shrinked log2FoldChange is -0.80 +- 0.15, so group2 have baseMean of 174.1 (100*(1/2^0.80)), and SE of 100*(1/2^-0.65)=156.9 to 100*(1/2^-0.95)=193.2"? I hope this is correct ... Sorry if misunderstanding.

ADD REPLY • link 6.7 years ago blastzone.heimerdinger ▴ 20

1

Entering edit mode

Sure that looks right for translating the LFC point estimate and SE to the count scale.

ADD REPLY • link 6.7 years ago Michael Love 43k