Normalization: merging Illumina GA and HiSeq RNA-seq datasets for DE analysis

[tejajo]

Hi,
I have two sets of samples sequenced using Illumina GA(II) and Illumina HiSeq. I would like to merge above two datasets so that I have more power to my differential expression analysis between two biological groups.

I am wondering whether the technical differences between two technologies (as shown in "Evaluation of genomic high-throughput sequencing data generated on Illumina HiSeq and Genome Analyzer systems") can be minimized or made comparable by a normal MDS scaling following a normalization by library sizes.

What are you opinions about that ?

Thanks in advance.

[Simon Anders]

Not sure how you could get normalization factors from MDS.

The better idea might be to fit an additional covariate. In the DESeq vignette, for example, we combine data from single-end and paired-end libraries. As we have both library types among both the control and the treatment samples, we can use a two-way ANOVA. You can follow exactly the same scheme as in the vignette, just by putting your two sequencing platforms instead of the two library types.