Library quantification quandary...

[axr624]

Hello! First time poster, although have lurked for a while.

I have looked through the site for advice on library quantification, and understand that ideally a combination of Qubit, Bioanalyser and qPCR should be used.

However, I am trying to use the MiSeq to generate reads out of transposons into neighbouring genome (TraDIS), and i'm not sure if the same constraints apply to my library prep in comparison to the majority of other applications.

Firstly, in TraDIS the number of reads is more important (to an extent) than the length of the reads. As such, is tightly defining the fragment length of my library as important as it is for other applications? Does having a wider range of fragment lengths on the flow cell affect cluster generation/PF/read quality? An important consideration here is that I would like to avoid any bias introduction through stringent size selection.

Secondly, assuming a wider fragment range in the library, would qPCR be the best option for quantification? Correlating Qubit concentrations and Bioanalyser traces becomes more problematic with a wider range library; would qPCR totally solve this issue?

As may be evident, I am still working through the library prep to finalise a standardised, reliable protocol. There are other aspects I need to look into, but I think it makes sense to iron out the basis of experiment. Any and all opinions are appreciated, and apologies for the long post!

[axr624]

As a follow up, I have used a non size selected library in two runs. The first run had 339k/mm2, 85% PF, 5.3m reads. The second, with twice the amount of library, achieved 414k/mm2, 84% PF, 6.2m reads. All other conditions were equal between runs...

[TonyBrooks]

You would still need to know what size your library is in order to correctly quantify.
Using the Qubit only measures the mass of dsDNA in the library. It tells you nothing about the number of DNA molecules (molarity) which is needed for sequencing. For a fixed amount of DNA, short DNA molecules will have a higher molarity than longer DNA molecules as their molecular weight will be smaller.
Similarly, if you qPCR you'll need to normalise the fluorescence values based on fragment length (assuming you're using a SYBR assay). Longer molecules = more SYBR fluorescence.
The problem comes in picking a correct bp value for these normalisations. Ideally, you'd like a nice tight library with not much size variation and no adapter dimer but this isn't always possible.

[rthornton4]

Interesting point TonyBrooks!! We were just debating this earlier in our lab. We are a Core facility that offers Illumina NGS. We will sequence 'user-prepped' libraries once they have been through our QC process which includes picogreen, BioA trace and KAPA qPCR. I just completed analysis on 12 user-prepped Agilent HaloPlex libraries. The library fragments span a range from 180bp to around 600 bp and it's not a smooth curve on the trace either. With this broad span of fragment sizes would you simply look at the 'Region Table' tab on the BioA trace and take the average? It does look like the average is calculated correctly but how would you normalize for the broad range of library sizes to determine the nanomolar concentration? It's an interesting question and I would love to hear opinions on this. Thanks!!

[SNPsaurus]

With a broad range of size, taking the average length will give problems. Say you have a distribution of insert sizes from 100-500. The average is 300. But it is not the case that the smaller fragments and larger fragments will balance out. There are 3x more fragments at 100bp compared to 300, while only 3/5s fewer fragments at 500bp compared to 300. Something like the geometric mean would be more appropriate (224 bp). Of course, having spikes and non-normal distributions adds complications!

Some fragment length analyzers do give regional molarity or calculate more accurately the molarity of a series of peaks.