DSN vs polyA protocol publications?

[Jeltje]

Hey all

Has anyone seen any publications in which the doube strand nuclease method of normalisation is compared to polyA selection? I'm curious if there are any differences in read levels for transcripts you'd expect in both (mRNAs of low to medium expression levels).

Thanks!

[SeqAA]

I'm sorry I cannot cite specifically, but DSN will reduce representation of everything. So depending on the sample, you cannot use the read counts reliably for further analysis.

[kmcarr]

Not a publication but a presentation from Illumina (download). It covers several topics on transcriptome sequencing; slides 35-36 cover DSN normalization. Specifically look at slides 44 and 46. They show from Illumina's studies that DSN normalization has little effect on the abundance of most mRNAs and the fold changes as measured by RNA-Seq using DSN normalized samples match extremely well to fold changes measured with qRT-PCR.

[Jeltje]

Thanks!
Illumina is, however, not completely unbiased. I was hoping to find some external validation.

[Camg]

mRNA-Seq vs DSN library gene expression variation

Hi,

I've had some Illumina sequencing done where my library was split so that part was used for the standard mRNA-Seq protocol and the other part (about 1/3 of the material) was used for a pilot run of the DSN normalization protocol.

I ran the data through Tophat and Cufflinks, and made a scatterplot (attached) of the expression levels (log transformed) of all the genes.
It looks like there is a decent correlation (Pearson corr=0.624) with some obvious variation. You can also see that some genes that have expression values in one sample were not detected in the other = the points that sit along the axes.
However, this probably isn't the best representation because this is based on quite a small amount of sequence data. So there might be less variation if each library was sequenced more deeply.

Hope this helps.

Cam

[cjschneider]

DSN vs polyA for transcript profiling

I think the jury is still out on this question. DSN will not necessarily decrease the frequency of all transcripts but is expected to decrease the most abundant RNAs in the sample. mRNAs are typically at low frequency in the total RNA pool so DSN treatment of total RNA should have little effect on relative mRNA transcript counts.

The results in the Illumina powerpoint presentation suggest that fold change estimated by transcript abundance from both DSN and poly-A preps have very strong correlations with qPCR estimated fold change (slides 17 and 46; r value not reported for regression in slide 17 but r = 0.947 in slide 46). If both methods have similar r values and there is no bias (i.e., abundant mRNA transcripts as measured by qPCR are not more often reduced in DSN libraries than in poly-A libraries) then it may be justifiable to use DSN as a library prep method for transcript profiling.

I am sure that many people would like to know about publications that compare poly-A preps and DSN preps against each other and/or against qPCR estimates of transcript abundance. Surely, someone has done those experiments. If so, please let us know!

[ester]

Now it works fine.
Thanks a lot,
Ester