Can I combine RPM values of RNA datasets from tow different databases?

Question

[akshay_ware]

I have RPM values of RNAs (eg.blood) from 2 different databases. Can I combine the blood sample RPM data from two different databases?

Answer 1

[ATpoint]

Technically you can, but you definitely should not.

1) You have no certainty that the processing of those data is identicalin terms of quantification and normalization.

2) Even if 1) was true and they were processed identically studies from different databases and laboratories are strongly confounded by the sequencing regime, kits used, lab protocols plus the uneven data processing. In my experience data from the same cell type will strongly cluster by study rather than cell type if coming from different sources.

3) If you even combine data then do it based on raw counts followed by joint normalization of all samples. Then explore if there are batch effects, e.g. using PCA on log2-transformed normalized counts that need to be addressed.

Here, better perform analysis on the databases separately and then do meta-analysis.

Answer 2

[WouterDeCoster]

No, you can't and you absolutely shouldn't.