[pmiguel]

Okay, I think you are getting ahead of yourself here. The simplest explanation for the result you posted initially is that your RNA is degraded. But it looks strange enough that it would be worth rerunning it. But for TruSeq DNAse treatment is recommended, so you might as well do that first -- if you clean everything up with a column. Then you can take a look at it on a chip again.

But, the most likely outcome is: degraded RNA, start again.

One step off the path as designed by Illumina, I'm willing to chip in suggestions. Two steps -- okay I'm still listening, not sure anything I say is going to be of much help. But you are 3 steps out now and on your own.

Let us know if it works, but you (1) have circumvented the normal QC step by declaring its failure was the result of genomic DNA in your sample -- which would not be there is you had not (2)also skipped a step asked for by the protocol by not DNAsing and now you (3)think that the fragmentation step should be sufficient to inactivate any DNAse still around after the polyA+ isolation. So, I can't say it will fail, but I don't like your chances.

The big problem is that the most likely failure mode, high 3' bias caused by low intactness of your total RNA, will not be obvious until you complete mapping back to a reference genome.
--
Phillip

[eab]

One more thing to factor in: these are irreplaceable human samples. There is no starting again.

So...I need to squeeze the best possible data out of them. If these data don't live up to my wildest dreams, I can accept that. 3' bias? Not the end of the world. I gain nothing by throwing the samples away.

Another consideration I have not made totally clear: there is very little RNA here. I do not like the idea of column purifying for that reason.

Definitely a mistake not to do on-column DNAse digestion, no argument from me there. I use RNAzol for my extractions, and this kind of contamination has not been a problem for me before. In this case, I have been trying to help a colleague make libraries from cells sorted into RLT buffer. Should it come up again, I shall know that the DNAse treatment is mandatory.

But, back to these samples. To me, there is just no way that a hump of material between two intact-looking rRNA peaks can be degraded RNA. If degraded material is going to pile up in a single hump, that better be a low-MW hump, not several Kb in length. I am sufficiently convinced by 1) the absence of a solid alternative explanation and 2) the similarity to the pics I sent out previously (see http://www.chem-agilent.com/pdf/5989-0991EN.pdf, especially page 3, item B) that I am going to proceed as if it's genomic DNA. And yes, I do think that the 95-degree, 6-min fragmentation step should be sufficient to kill any DNAse that manages to persist through two rounds of dynabeads purification (which to my mind isn't so different from a column purification, anyway).

But of course I could be wrong. The plan is to pool aliquots of residual material I had frozen separately for the Bio-A, and make one single library according to my made-up method. I will post results either way.

Thanks all for your advice,
Eli

[kerplunk412]

Dynabeads and DNAse?

I don't think it is a good idea to have active DNAse during the dynabeads procedure- aren't the oligos made of DNA? If so this will ensure that you don't recover any RNA, as the DNAse will degrade the oligos so there will be nothing for your RNA to hybridize to. Also, you run the risk of seeing lots of oligo dT in your final library. Maybe I missed something?

[eab]

nope, don't think you missed anything. yeah, the "d" in "oligo-dT Dynabeads" is, like, the same d in DNA. embarrassing.

thank you!