[BioC] comBat or PCA to adjust for hidden variables in a high-throughput qPCR experiment with only 120 genes measured

[shirley zhang]

Dear List,

We have performed a high-throughput rtPCR experiment for selected 120
genes on 2,000 human subjects using TaqMan platform. There are two
batches/runs, and three housekeeping genes used across all
plates/runs.

However, since there are about 2,000 blood samples which were NOT
drawn at the same time from individuals, and RNA were NOT extracted at
the same time.

In genome-wide microarray data which include all 20,000 human genes ,
we could apply comBat or PCA to adjust for hidden variables. However,
in our rtPCR data, there are only 120 genes measured on 2,000 samples,
can anybody advise us whether comBat or PCA method is still applicable
here? If not, do you have any suggestions for controlling for the
latent variables?

Please keep in mind that >90% of these 120 genes are highly or
moderately expressed in blood samples.

Many thanks,
Shirley