Rna-Seq Data For Cd34+ Cells Or An Alternative Approach To Compare My Expression Set

Question

[Dataminer]

Hi!

I am looking for a RNA-seq data on CD34+ cells (human), to compare the expression of genes in my treatment sample and identify differentially regulated genes.

I know microarray data for CD34+ cells is available but I do not want to compound my problems by making a cross platform comparison.

Or am I missing something here when I am saying/thinking that a comparision of expression data from microarray with RNA-seq will not yield differentially regulated genes?

Or what will be your approach to solve this problem?

I have a RNA-seq profile for my treatment sample and microarray profile for my control sample.

I know it is a tricky one and many will down vote this question (It usually happens with my questions, don't know why? for example sometimes people don't understand what is peak file and they down vote the question).

Anyway, I still hope for an answer.

Thank you Note: GEO and Oncomine have been querried.

Answer 1

[David Quigley]

The first point is that your experimental design is not helping. If you want to do a good job of comparing treated and untreated, you need to measure them the same way. You also don't say whether you have replicates of your treated and untreated samples; if not, I suggest you just stop right there and do the experiment in a rigorous way. In order to make a strong case, what you want is to say, "in our own hands, we measured CD34+ cells from three different people and then treated three people with drug and extracted CD34+ cells in the same way (same FACS sorting protocol, etc) and measured with RNA-seq and saw this difference". Otherwise, just the fact that two groups used different gates for the FACS is enough to make me doubt that "CD34+" really means anything consistent.

That said, there's no reason why, in principle, methods used to contrast of RNA-seq with microarray data should be different from methods use to contrast two different microarray platforms. In general, one could get a single call per gene per sample for each platform (e.g. collapse probes on the microarray, summarize by gene on RNA-seq) and find the intersection of gene lists. You could try a comparison of relative ranks of each gene within a given sample, since it won't make sense to compare directly (for example) RPKM on RNA-seq vs. Log2 of expression on microarrays.

[Dataminer]

Point taken, thank you.

Answer 2

[Sebastian Boegel]

I dont know if this answer still helps you. I found 4CD34+ RNA-Seq datasets:

http://www.ncbi.nlm.nih.gov/sra?term=SRX037948

http://www.ncbi.nlm.nih.gov/sra?term=SRX135562

http://www.ncbi.nlm.nih.gov/sra?term=SRX174326 (30 GB!!!)

http://www.ncbi.nlm.nih.gov/sra?term=SRX105932 (this PE, although GEO shows only the SE sign).

I hop it helps

[Dataminer]

Thank You :) : )