Hi @ all,
I'm comparing MicroArray based expression data with RNASeq expression data. For this I have data of two fungi strains, one named B36 the other named N402, each with 3 replicates.

I observed quite low correlation between the the RNASeq samples. In the plot you can see the Pearson correlation of the normalized read counts and MA intensities (not log2 transformed).

The correlation between the MicroArray samples is quite high but between RNASeq samples it is quite low. The N402 RNASeq sample shows an even higher correlation to both MicroArray samples than to B36 RNASeq sample.

I used the same mapping and counting procedure for both samples (STAR mapper + R package bamsignals). I observed a similar number of mapped reads and a similar sum of reads counts in both samples. The correlation between the RNASeq B36 and N402 replicates is near to 1. The fastqc report is quite good for all samples.

Does anyone have an explanation for this? Is my RNASeq B36 data just rubbish? Are there any other quality control steps I can do?
Or do you think the data is ok and RNASeq is much more sensitive to expression changes?

Would be very nice if someone could help :)