Thursday, 6 August 2015

Quantification and equimolar pooling of NGS libraries

Library preparation for next generation sequencing is becoming easier with the quality of kits and protocols improving substantially in the past few years. With the price of NGS decreasing, we are finding that our throughput is increasing, both in terms of the number of experiments as well as the average size of these experiments. With this in mind, the ability to accurately pool barcoded libraries in equimolar ratios (also called "balancing") is even more critical. Accurate quantification is thus vital. There are several ways to quantify library DNA:
  • Qubit fluorometer. Gives very accurate concentrations in nanogram per microlitre, but agnostic of fragment size distribution.
  • Bioanalyzer. Gives accurate readings of fragment size, but is only fairly accurate in terms of concentrations of each fragment
  • Quantitative PCR. Most accurate, but expensive and most time consuming.
  • NanoDrop UV-Spec. Easiest method but least accurate. Not recommended.
So I'll lead you through my favourite method for equilmolar library pooling that only uses a bioanalyzer, specifically the Shimadzu MultiNA.

When you elute the DNA off the column or magnetic beads, the first thing you need to do is look at the bioanalyzer profile. Here is an example of a bioanalyzer result for 23 samples (left) and a detailed look at the first sample (right). The MultiNA software also kindly calculates the molarity of the peaks detected, showing that sample 1 is 338.53 nM.

As you can see in the left panel above, is that the library concentrations do vary quite a lot. To demonstrate this more clearly, I present the concentration value of the main library peak for the whole 30 samples in the same experiment in the below graph. You can see that the libraries vary in concentration from 450nM down to 50 nM, meaning the pooling step will be prone to a high errors if the MultiNA readings are just a bit inaccurate.

As the accuracy of MultiNA is not that good over this wide range of concentrations values, its better to dilute the libraries down to a common concentration first and then quantify again on MultiNA. As you can see below, the variability is much improved as compared to the above graph, but not yet perfect.

This second measurement reading is used to calculate the volumes required to generate the library pools. The data yield after quality trimming shown below demonstrates that the balancing has improved again, and the three pools, each containing 10 barcoded samples are well balanced.

Prior to clustering the libraries, I'd recommend running the pooled library on the MultiNA, followed by dilution to 10 nM and then final quantification with either MultiNA or Qubit. The approach I use is fairly cheap as MultiNA runs cost as little as 10c per sample and is not very labour intensive compared to qPCR methods.