Question: ComBat before or after Probe filtering ????
gravatar for Sathya
6.2 years ago by
Sathya30 wrote:


  I  am planning to preprocess 40 microarray series comprising of 1200 samples. For all these series, mRNAs were quantified using same platform "GPL570". I thought of using ComBat for batch effect correction prior to differential analysis. But, I dont know  at which step that I need to use comBat.

Which is the correct procedure?

Background correction, Normalization , filtering the probes based on present and absent calls (panp package in R) and combat


Backgroung correction, Normalization, batch effect removal using combat and filtering the probes based on present and absent calls (panp package in R).

There is no defined method and Each paper follows different methodology. Please let me know.

Thanking you for your time.




ADD COMMENTlink modified 5.4 years ago by brendan.innes10 • written 6.2 years ago by Sathya30
gravatar for brendan.innes
5.4 years ago by
brendan.innes10 wrote:

I was under the impression that due to the Bayesian nature of ComBat, it worked best with as much information re: distribution as possible, therefore should be done before filtering.  

Side note: filtering by variance seems to be a popular alternative to detection calls.  See bioconductor package "genefilter".
I like it because I found that panp doesn't play nicely with the updated CDFs available at brainarray.

ADD COMMENTlink written 5.4 years ago by brendan.innes10
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1591 users visited in the last hour