Comparing DEG result from different experiment

Question

[bharata1803]

Hello all,

So I have a simple question. Suppose I download 2 or more dataset from NCBI GEO. Those dataset come from different experiment especially the NGS machine. For example one of them use Illumina and others use Ion Torrent.

My first question is, if I calculated each dataset separately for DEG, using EdgeR, Limma/Voom, or DEseq2, are these log fold change comparable?

For example, Dataset 1 comes from Illumina Hiseq and I have logfoldchange from DESeq2. ANother Dataset 2 comes from IonTorrent and I have logfoldchange from DESeq2.

My second question is, what if I calculate the DEG using different method? One of them using DESeq2 and other use Limma/voom?

What should I consider?

Thank you.

[Lluís R.]

If they are properly normalized and are the same comparisons on the same biological and experimental background yes. Different methods should agree on the magnitude of the change if not in the value itself, apply common sense.

[bharata1803]

Thank you, I just want to make sure it is make sense. I have several times compare DEG analysis using DESeq2 and Limma and edgeR with same dataset and they produce similar result.

[Sean Davis]

The literature might be helpful:

• https://doi.org/10.1186/s12864-017-4011-0
• https://doi.org/10.1038/nbt.3000