Remove batch effects using ComBat


Hello from GenomeSpace! In this video step we will attempt to remove
batch effects from our gene expression dataset using the ComBat module in GenePattern. From the Modules tab search or browse for
the “ComBat” module. ComBat requires continuous gene expression
data such as microarray intensity. Thus, we previously converted our discrete RNA-Seq
count data to the continuous log counts-per-million unit using the PreprocessReadCounts module.
We access the outputs from the PreprocessReadCounts module under the Jobs tab. Click and drag
the gene expression file, in this case “GSE63412_GSE44229.logcpm.gct”, to the input file field. ComBat also requires a sample information
file that annotates the samples for the batch covariates as well as covariates of interest.
From the GenomeSpace tab, click and drag a sample info file such as “GSE63412_GSE44229.sample.info.txt”
to the sample info file field. We will be using the “parametric” prior method.
If the parametric estimates are poor, we can rerun the job using the “non-parametric” method. Click “Run” to submit the job and wait for
it to finish.

Leave a Reply

Your email address will not be published. Required fields are marked *