Question

Design for introducing batch effect in DESeq2

0

Entering edit mode

Arindam ▴ 80

@ag1805x-15215

Last seen 4 weeks ago

University of Eastern Finland

I have RNAseq data taken from two different ENA project IDs (PrjA & PrjB). Both PrjA & PrjB contains test and control samples. PrjA was PE 50bp sequencing while PrjB was SE 100bp. Is it possible we can combine the data to identify DEG between test and control? Would batch effect removal by Combat be a better option or use the design = project+condition in DESeq2?

deseq2 combat batch effect rna-seq edger • 1.3k views

ADD COMMENT • link updated 5.2 years ago by swbarnes2 ★ 1.4k • written 5.2 years ago by Arindam ▴ 80

0

Entering edit mode

If you DO know the batches, then what swbarnes2 recommended is the best option. If you don't know then you could use sva.

In your case, however, I don't think that it's wise to just combine the data because it seems that you use different platforms for the sequencing. Could you show us the MDS plot?

ADD REPLY • link 5.2 years ago mikhael.manurung ▴ 280

score 1 · Answer 1 · 2019-08-30

1

Entering edit mode

swbarnes2 ★ 1.4k

@swbarnes2-14086

Last seen 3 days ago

San Diego

Do what the DESeq software recommends. Give DESeq raw counts, not some kind of batch corrected, and include batch as a factor in the design.

ADD COMMENT • link 5.2 years ago swbarnes2 ★ 1.4k

1

Entering edit mode

You can do that even when the data came from two different projects? Seems like the batch issue in this case is not simply multiple runs of the same instrument but other things are different as well.

ADD REPLY • link 5.2 years ago mikhael.manurung ▴ 280