Troubleshooting Manual NGS Library Prep

Large-scale pharma and clinical labs use liquid handlers for routine NGS library preparation, and so avoid many of the human errors associated with manually preparing a library. But many researchers do not have the volume, or the budget, for automation to make sense for their labs.

Here we look at some things that can go wrong during manual NGS library prep, how to avoid them (hint: paying close attention to procedural instructions is key), and (in rare instances) what can be done to salvage the prep.

Importance of following protocols carefully

Biocompare consulted with several seasoned veterans for their expert perspectives and advice. They all noted that perhaps the most frequent causes of errors are due to not following procedural instructions at every step, including not paying adequate attention to, or not heeding, what is supposed to be done and how — often unintentionally on the part of lab personnel. This can take many forms, from not following instructions, to becoming distracted, to relying on outdated protocols, to working on autopilot. Here are a few select examples.

Unreliable reagents can ruin a prep. Make sure they have been stored properly, are handled properly (kept on ice, for instance), and aren’t expired. Make up 70% or 80% ethanol fresh and at the correct concentration—it can evaporate quickly, and too much water can elute your DNA from the beads earlier than desired. Follow instructions as to whether to mix by pipetting up and down or vortexing, and for how long—keeping in mind that some enzymes may be sensitive to overmixing.

If you’re doing a new protocol that’s similar to one you are familiar with, it’s easy to fall into habits. RNA preps, for example, require a DNAse step not found in DNA preps—otherwise the final NGS results will be overwhelmed by DNA sequences. Or sometimes you can confuse similar steps in a repetitive protocol, like when you inadventently take the supernatant and discard the beads, or vice versa. There are a couple of potential remedies for the latter, though. You can put the discard into a “waste plate,” to be retrieved just in case you realize your mistake in time. A more preventive measure would be to highlight or circle the critical steps in the protocol to help you pay more attention.

Sometimes a reagent fails to make it into a tube, or is dispensed twice, when running a lengthy protocol with multiple samples. To help prevent this from happening, try using a master mix to help minimize the risk. Another tip is to use your box of pipette tips as a map to align the samples to the tips used in the tip box. This will help confirm that reagent was added to each tube. And if you’re not sure whether or not you added reagent to a tube (or put it in twice), you can measure the volume and compensate accordingly (it pays to keep extra beads around just in case).

Mixing up barcodes is another oft-cited mistake that can be prevented. Plan ahead, including whether to opt for a single-use index, combinatorial dual index, or a unique dual index. Double check the wells you took the barcodes from. And record everything (including mistakes) immediately.

Quality (and quantity) control

Using low-quality DNA can result in a low-quality library, because what goes into a prep is as critical to a good library as the prep itself. Run both the input nucleic acid and the output library on a Bioanalyzer or similar-type instrument to determine its integrity. Use more of samples of low quality in order to get the required amount of DNA for sequencing.

If there are too many adapter- or primer dimers in the library, it won’t sequence well—the BioAnalyzer should give this information. Make sure to clean up the sample, maybe even twice.

It’s important to know the concentration of your DNA (and not everyone has access to a BioAnalyzer). A spectrophotometer (such as NanoDrop) is good for determining nucleic acid purity, but doesn’t give a very accurate measure of usable DNA. A fluorimeter (like Qubit), on the other hand, gives a good reading of how much double-stranded DNA is in the sample, but says little about purity or integrity. qPCR can quantitate how much amplifiable material is in a sample. It’s recommended to do both Qubit and qPCR, except for amplicon libraries, which can rely on just qPCR.

Pipetting

Manual pipetting can be a major source of the human error responsible for low-quality libraries. Pipettes may not be properly calibrated. Tips may not be attached securely and therefore not make a good seal. You may not be aspirating fully, or dispensing fully. This can happen one time, or even systemically. Or you may inadvertently skip a tube. To help prevent the latter, use strip tubes with separate lids and close each lid once the mix is added.

A better option might be to use a multi-channel pipettor. These can save a lot of time and effort, and can prevent tubes or wells being missed. But they don’t always aspirate and dispense evenly. It’s critical to calibrate them, to visually inspect the volume drawn across all the tips, and to make sure that everything has been dispensed.

To end on a high note … not all errors doom the prep to failure. In the end, if you find that there is not enough material to load onto the sequencer, the prep may still be salvageable. There is usually a universal amplification step at the end of the protocol using the common adapters on the ends of all the library molecules, and if so, it should be possible to go back and do some extra amplification.

Many thanks

The following experts’ insights powered this article:

Mark Dasenko, Senior Faculty Research Assistant II, Center for Quantitative Life Sciences, Oregon State University; Giorgia Del Vecchio, Assistant Project Scientist, University of California, Los Angeles; Isabel Gautreau, Senior Tech Support Scientist, New England Biolabs; Adam Harris, Director of R&D, Clinical Next Generation Sequencing, Thermo Fisher Scientific; Vanessa Schmid, Lab Manager, Next Generation Sequencing Core, UT Southwestern Medical Center; Sanda Zolj, Technical Support Scientist I, New England Biolabs