Counts plot does not match fold change between groups

Question

[juara]

Hello all,

I have patient data that are grouped based on a characteristic. I ran DESeq2 between the two groups following the tutorial. Everything seems fine but when I plot one of the top differentially regulated genes using plotCounts(), it seems the difference is driven by few outliers. Here is one example that DESeq2 outputs more than 6 log2FoldChange that is highly significant:

I would appreciate if you can help me understand this issue. My feeling is that the counts are not normalized so it may not show the difference?

Thanks

Answer 1

[swbarnes2]

Driven by outliers or not, (you really want to say that 3/7 samples are outliers?) the two sets of samples pretty clearly have different means, so the software was right to flag that.

Answer 2

[jordi.planells]

Have all the sequencing libraries been processed equally? Do they come from the same run? Maybe you have a batch issue there.