DESeq2 outliers-should I be concerned?

Question

[bioinfo]

Hello,

I ran DESeq2 for 2 projects recently. One of them ran fine but the other gave me this message:

-- replacing outliers and refitting for 35 genes  
-- DESeq argument 'minReplicatesForReplace' = 7  
-- original counts are preserved in counts(dds)

Should I be concerned? I have 35772 genes in total and DESeq2 completed without error. The only weird thing was that message.

Thank you

Answer 1

[ATpoint]

It means that DESeq2 has replaced outliers in 35 genes because they had a large Cook's distance and you had more than 7 replicates so a certain confidence could be applied that the outlier is indeed an outlier rather than a potential genuine effect or group spread. The replacement tries to denoise the fit. Nothing to worry about I would say, it's an established built-in function that exists for a long time. You can turn it off, see vignette, if you feel like so.

[bioinfo]

Thank you so much for the explanation!