Limma biological replicates only in one group
1
0
Entering edit mode
obarton ▴ 10
@obarton-11523
Last seen 8.4 years ago

Hi,

I'm performing a differential analysis using limma between control samples and samples from two related diseases from approximately 100 patients in each group. The experiment is simple in that we are trying to find transcripts differentially expressed in each disease compared to control samples.

The diseases tend to occur more frequently at different body sites on the skin, so in the design of the experiment, we have collected two samples for each control sample. In some cases we have collected 3 control samples and all of the diseased samples are from independent patients. The numbers of samples look like this.

       arm back leg scalp
  CTRL  14   99 100     0
  D1    37    2  44     0
  D2    10   90   1    19

I'm currently using limma whilst blocking for body site with the contrasts D1-CTRL and D2-CTRL however I'm thinking that treating replicate control samples from different body sites as independent samples is incorrect and that I am underestimating the variance in the control group. Could anyone suggest the best approach to deal with this? Would adding the replicates as a random effect using duplicateCorrelation() be appropriate in this instance? Ideally I’d like to keep all control samples in the design so that I'm comparing against a common control baseline.

 

Cheers,

Owen

limma microarray • 1.1k views
ADD COMMENT
0
Entering edit mode

It isn't clear what you mean when you say "We have collected two samples for each control sample." It seems that you are using the word "sample" for two different things. Do you mean that you took two tissue samples from each part of the body for each healthy individual? If not, could you explain more explicitly what you have done?

ADD REPLY
0
Entering edit mode

Apologies, that was a very confusing mistake. I meant that we collected two samples (one at each body site) for each healthy individual. 

ADD REPLY
0
Entering edit mode

OK, that's clear now.

ADD REPLY
1
Entering edit mode
@gordon-smyth
Last seen 22 hours ago
WEHI, Melbourne, Australia

Yes, using duplicateCorrelation() to correlate the samples from the same individual would be very appropriate in this situation.

ADD COMMENT
0
Entering edit mode

Great thanks

ADD REPLY

Login before adding your answer.

Traffic: 813 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

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

Powered by the version 2.3.6