Question

DESeq2 with 2 biological replicates: if the replicates are not compatible for some genes

0

Entering edit mode

f_rahmdani • 0

@f_rahmdani-23913

Last seen 12 months ago

Canada

I did DE analysis on 80 libraries with DESeq2, the number of biological replicates is 2 (40 samples). For some genes, there is a big difference between read counts for 2 replicates, for example r1:2 and r2:2436. If the number of replicates was more than 2, these cases were removed as an outlier. How to solve it with 2 biological replicates? Thanks

deseq2 • 1.0k views

ADD COMMENT • link updated 3.8 years ago by Michael Love 41k • written 3.8 years ago by f_rahmdani • 0

score 2 · Answer 1 · 2020-07-27

2

Entering edit mode

Michael Love 41k

@mikelove

Last seen 18 hours ago

United States

There is no great answer that statistical methods can provide here.

This reason is exactly what I say to people who propose to perform n=2 experiments: "What will you do if the two replicates are in total disagreement? How will you know if either or both experiments had a technical issue?"

ADD COMMENT • link 3.8 years ago Michael Love 41k

0

Entering edit mode

Thanks a lot Michael for the quick response! You are absolutely right; but is there any way to eliminate both replicates in these cases?

ADD REPLY • link 3.8 years ago f_rahmdani • 0

1

Entering edit mode

I don’t have any recommendations other than using QC metrics per sample, eg MultiQC