small library size (HiSeq)

[Robby]

Hello everyone,

we often have the the problem, that we have libraries with small size (ie small RNA), but we don't have enough samples to sequence a complete flowcell. So we were wondering, if it is possible to sequence the libraries with small size on a 100bp HiSeq run and use afterwards just the first 30-40bp. Are there any problems with that (sequencing or cBot) or would a small library size on one lane influence for example other lanes? Has anyone experience with that or any recommendations? I would be happy for all advice.
Thanks Robby

[simonandrews]

I can't see why that would be a problem (if you're prepared to pay the extra cost of a longer run). You'll just end up sequencing a load of adapter sequence, and plenty of people have done that (mostly unintentionally!) before.

You shouldn't have any effect on the other lanes as long as your lane hasn't been specified as the control lane for the run.

[Robby]

OK, thanks for your information.

We know, that we have more costs for that lane due to the longer reads, but it is still cheaper for us than a complete run with some empty lanes.

[SeqAA]

As long as it's the minority of the flowcell, this should work. You run the risk of saturating the imager when you start sequencing the through to the adapter. Low complexity would be my concern.

[simonandrews]

If the low complexity comes after the sequence you care about then it shouldn't matter too much. If your inserts are variably sized then it won't matter at all. The only problem you might face is that the sequences might get kicked by the QC filter if the later base calls get too rubbish, but you could just choose to ignore the filter and take the raw calls in that case.

In any event it shouldn't affect anything outside the current lane.

[Robby]

@SeqAA: What do you mean with "risk of saturating the imager"? We would sequence 2-3 complete lanes with small RNA.In the other lanes we would sequence "normal" libraries. If we are behind the second adapter sequence, we will of course receive just N's. But do you think, that that influences the first bp? Or do you mean that this influences the other lanes? So for which part would you expect low complexity?