Question: Expected Correlation Between Exon Array And Rna-Seq
gravatar for polarise
7.0 years ago by
Galway, Ireland
polarise380 wrote:

I've carried out an experiment involving both of these platforms and the Spearman cross-correlation between them for gene expression (using Affy Power Tools for exon array and Cufflinks for RNA-Seq) are between 0.55 and 0.60. I've had a look at several papers with similar results and the Spearman correlations for quite a number tend to be in the ~0.8 and above range.

Question: If you have published work with such data what was your Spearman correlation?

Thanks, Paul

rna correlation microarray • 2.9k views
ADD COMMENTlink written 7.0 years ago by polarise380
gravatar for Pablo Marin-Garcia
7.0 years ago by
Pablo Marin-Garcia1.8k wrote:

I assume that you have seen John Marioni's paper RNA-seq: An assessment of technical reproducibility and comparison with gene expression arrays

Comparison of results across technologies

As a first step to comparing the sequence and array data, we compared the number of sequence reads mapped to each gene with the corresponding (normalized) absolute intensities from the array (Fig. 3). Reassuringly, these two independent measures of transcript abundance are highly correlated (Spearman correlation = 0.73 for liver, 0.75 for kidney). Interestingly, where results from the two technologies differ, it is generally where the array intensities are large and the sequence counts small; a pattern that might be explained by probe-specific background hybridization on the array.

ADD COMMENTlink written 7.0 years ago by Pablo Marin-Garcia1.8k

Yes, I've had a look at those papers.

ADD REPLYlink written 7.0 years ago by polarise380
gravatar for Philippe
7.0 years ago by
Barcelona, Spain.
Philippe1.9k wrote:


I would say a spearman rho between 0.55 and 0.6 is somehow expected. When comparing microarrays and RNA-Seq data from a same biological sample I have a spearman rho around 0.60 or 0.65 (Agilent arrays vs lab-made pipeline for RNA-Seq data for mammalian samples). Also, if you look at this article they describe a correlation coefficient lower than 0.6 (and even quite close to 0 from lowly or highly expressed genes - data from yeast - See Figure).

This shows that having a correlation coefficient around 0.5 or 0.6 for such comparison is not unexpected. I guess higher coefficients might be obtained by comparing subset of relevant genes or using different mapping/quantification procedures for RNA-Seq.

ADD COMMENTlink written 7.0 years ago by Philippe1.9k

Thanks for that. I hope I get many more responses to see what others think.

ADD REPLYlink written 7.0 years ago by polarise380
gravatar for Malachi Griffith
7.0 years ago by
Washington University School of Medicine, St. Louis, USA
Malachi Griffith17k wrote:

Another paper with a variety of comparisons between Affymetrix Exon arrays, custom NimbleGen arrays, and RNA-seq: Griffith, et al. Alternative expression analysis by RNA sequencing. Nature Methods. 2010 Oct;7(10):843-847.

Most of the relevant results are in the supplementary materials, in particular Supplementary Figures 9a-f. Spearman correlation at the gene expression level was 0.78. Supplementary Figure 10 describes the outcome of validation experiments involving RT-PCR, qPCR, and Sanger sequencing...

ADD COMMENTlink written 7.0 years ago by Malachi Griffith17k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 2123 users visited in the last hour