batch correction in DESeq2 > matrix not full rank > mitigate
Entering edit mode
MT ▴ 10
Last seen 3.3 years ago

Hello, I am a novice and I am interested in comparing two kind of cells coming from different RNA sequencing experiments published. The design would be something like the following:

sample    cell    batch
1    type1    1
2    type1    1
3    type1    1
4    type1    2
5    type1    2
6    type2    3
7    type2    3
8    type2    3
9    type2    4
10   type2    4

Now, naturally I cannot model del batch effect with ~ batch + cell because the batch is nested in the condition "cell". But the question remains, what is the best way to compare type 1 and type 2 taking into consideration and mitigating possibile batch effects?

Thanks for your help. Best, Marco

DESeq2 batch correction • 377 views
Entering edit mode
Last seen 7 hours ago
United States

The short answer is that you can't do that. There's no way to resolve the confounding.

Entering edit mode

Perhaps you can try surrogate variable analysis?

Entering edit mode

Surrogate variable analysis isn't magic, and won't fix confounding.

Think of it this way. If you had a study where you treated a group of patients with a drug and had a group of patients who were controls, but at the same time all of the treated patients were smokers and the controls were not, how would you ever know if any differences were due to the treatment or the smoking?

SVA is just a way to adjust out unobserved factors, but it won't help with completely confounded variables. There is no magic way to fix that.


Login before adding your answer.

Traffic: 283 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6