RNA-seq: Replicates, single-end, paired-end story

[pasta]

Hi,

In my lab they made a first RNA-seq run with single-ends, but because the coverage was a bit low they decided to make a new run on the same samples 6 months later, with the paired-end technology ...
The coverage with the Run #1 is 5-fold lower than with Run #2 (40 million reads). When I plot the 2 runs I obtain a fairly good correlation (r= 0.95).

So now I have the choice to work with the last run (paired-ends) but with NO replicates, or use runs #1 and #2 as replicates but with a mix of the paired-end/single-end technologies. I do not know what is the best solution here - do you have any idea ?

Thanks

[Simon Anders]

Have a look at the recently updated version of the DESeq vignette, where we use the data of Brooks et al. to demonstrate how to use GLMs to handle blocking factors such mixtures of different sequencing styles.

However, this probably does not apply to you. As you talk about only one sample, you don't want to compare anything anyway. And then, you might not even need replicates, depending on what you want to look for in your single sample. (And if you do need replicates, you are screwed anyway, as you don't have biological replicates.)

[pasta]

Thanks Simon for your answer.
We do have 2 conditions to compare:

condition 1 : Paired-end run + Single-end run
condition 2 : Paired-end run + Single-end run

You are right, since we do not have biological replicates but only technical replicates, I guess we can forget about this idea.