How high/low is library construction efficiency?

Question

[Phil S.]

Hi guys,

This is more of a wet lab related question but since it have might affected one or another of you I'll try my luck.

Does anyone of you have a reference or an idea of how efficient the library preparation (I know it might differ from protocol to protocol significantly) is? Some good estimates should give me a starting point.

Edit: To be more specific. What I mean is that from a fixed amount of DNA I can calculate how many molecules are in my sample. However, what I need to know is:

1. How many of those molecules actually end up getting caught by the sample preparation?
2. If this is possible to know: How much of the provided library is getting a place on the flowcell itself? (This might be the trickiest one since you can adjust with molarity of the library itself but for sake of simplicity assume there should be enough space for the whole library/ every molecule)?

Thanks in advance

Best
Phil

Answer 1

[harold.smith.tarheel]

1) Library construction is inefficient (<0.1%). Very rough back-of-the-envelope calculation: typical microgram-scale input represents ~10e12 molecules, whereas typical library complexity is ~10e8 molecules. I believe qPCR measurements of PCR-free libraries report similar inefficiency.

2) Clustering (flow cell capture) is much more efficient, on the order of 30% for HiSeq 2000 (based on potentially flawed memory; I calculated it once from the loading molarity, flow cell volume, and number of clusters obtained).

[Phil S.]

awesome thanks, that's enough info I need for a first shot ! (but btw. do you have any reference for that in literature?)

[harold.smith.tarheel]

No reference, just experience. But there are plenty of publications on NGS library complexity, and it's simple enough to calculate molarity from the input metrics (mass and size). The same is true for clustering, based on values obtained from Illumina's user guide/instrument specs.