Improving NGS Library Prep with an Open

System Liquid Handler

Matthew J. Nesbitt1 and Masoud M. Toloue2 1Aurora Biomed Inc, 1001 E. Pender St., Vancouver, BC V6A1W2 2Bioo Scientific, 3913 Todd Lane Suite 312 Austin, TX 78744 Correspondence should be addressed to M.T. (mtoloue@biooscientific.com) or M.N. (matt@aurorabiomed.com)

Growing next generation sequencing (NGS) throughput and new, sophisticated multiplexing

strategies have significantly increased the number of samples users can prepare. The de-

mand for reproducible and reliable data requires consistency between libraries. To achieve

this, the VERSA Mini NGLP, liquid handling workstation has been optimized for NGS library

preparation. Using NEXTflex™ library preparation technology in conjunction with the VERSA

Mini NGLP, high throughput functionality matched with superior enzyme performance out-

paces and improves consistency when compared to manual library preparation.

INTRODUCTION The plummeting cost of sequence acquisition that has

resulted from advances in next generation sequenc-

ing (NGS) technologies has led to its use in new set-

tings such as small research groups, contract research

organizations and the clinic1. The availability of new

benchtop NGS platforms such as the MiSeq from Illu-

mina and the Ion Torrent PGM from Life Technologies

is encouraging NGS adoption by these facilities. The

offer of low per base costs and short run times with

relatively small up-front instrument investments2 aids

the justification these groups use to introduce NGS to

their labs, and drives the eventual adoption of higher

capacity sequencers. Groups that traditionally out-

sourced sequencing to large science facilities have

been empowered, and can bring their projects in-

house. Widespread adoption of NGS is quickly ap-

proaching in spite of an underdeveloped field of liq-

uid handling workstations validated to complete the

associated library preparation protocols. Because of

the high number of sample manipulations needed for

this task, consistently reliable results depend on liq-

uid handling automation3.

VERSA Mini NGLP™

The delicate sample handling offered by the VERSA

Mini NGLP makes this automated liquid handling

workstation ideal for NGS library preparation proto-

cols. A single channel, dual syringe pipette manifold

that is paired with high resolution stepper motors

provides accurate liquid handling at variable speeds

over a large volume range. Reaction setup, incuba-

tions and sample cleanups are executed in a UV lamp,

HEPA filtered hood. The dispense-only ReagentDrop

clusters provide a quick and contamination-free

method of precipitating sample nucleic acids.

Flexible software that provides the user a high degree

of control over the mechanics of liquid handling is

provided. Varying aspiration and dispense speeds can

be set for individual operations to minimize sample

perturbation during magnetic bead-based cleanups.

New protocols can easily be automated, and users

can easily update the number of samples to be proc-

essed. The VERSA Mini NGLP provides an appropriate

automation solution for groups looking to leverage

the advantages of 3rd party reagent kits while intro-

ducing consistency between library preparations.

Protocol

Protocol

NEXTflex™ LIBRARY PREP KITS

Designed to prepare single, paired-end and multi-

plexed genomic, ChIP, mRNA, small RNA and reduced

representative bisulfite sequencing libraries, NEXTflex

(Bioo Scientific) improves on current library construc-

tion by using enhanced enzymatic master mixes and

flexible multiplexing options. NEXTflex enhanced

adapter ligation technology is one example. Using

specially designed ligases and an optimized reaction

buffer, the resulting higher number of ligation events

(adapters ligating to sample) with NEXTflex ligation

mix directly correlates with a higher degree of se-

quence diversity, better read numbers, coverage and

assembly.

NEXTflex kits are designed to be compatible with high

-throughput liquid handling automation by using

master mixed enzymes and buffers and gel-free size

selection. With growing sequencing throughput, us-

ers are increasingly able to multiplex or pool several

samples into a single sequencing reaction. To accom-

modate this capacity, NEXTflex DNA Barcodes (Bioo

Scientific), can be used to multiplex or simply tag

samples to monitor cross contamination with up to

96 unique indices. This significantly reduces sequenc-

ing associated costs, allows measurement of base

error rate, cross genomic studies, time course and

drug induced cellular experiments. The NEXTflex DNA

barcodes are available in 96 well plates enabling high

throughput automation.

NEXTflex and VERSA

Low cost automation of NGS library preparation with

the use of optimized library preparation reagents

offers flexibility for groups that leverage multiple NGS

technologies with varying workflows. Laboratories

with multiple NGS technologies can automate differ-

ent library preparation protocols with a single re-

agent provider on the same automated platform.

Currently, most validated automation options for

library preparation are closed systems that do not

accommodate a broad range of reagent kits.

The University of Arizona Genetics Core validated the

use of the VERSA Mini NGLP (Aurora Biomed Inc.)

with NEXTflex (Bioo Scientific) for NGS library prepa-

ration. As an open-platform workstation, the user is

no longer tied down to one library preparation kit.

This can drastically drops the cost per sample. In ad-

dition, the accompanying VERSAware user software

enables control over aspiration/dispensing speeds

and tip positioning that can be tailored to delicately

handle several types of genomic samples.

METHODS

DNA isolated from the Pocket Mouse (Chaetodipus

intermedius) was used as a template for library

preparation. Fragmentation was completed using a

Covaris S series instrument (Woburn, MA, USA) and

split into two equivalent volumes. One was processed

using the standard Illumina manual library prepara-

tion (manual library) used by the University of Ari-

zona Genetics Core. The other was processed in an

automated fashion (automated library) with the

VERSA Mini NGLP Workstation from Aurora Biomed

(Vancouver, BC). Enzymatic modifications of this sam-

ple and necessary purification steps were completed

using the NEXTflex DNA Sequencing Kit (cat# 5140-

01) from Bioo Scientific (Austin, TX).

Reaction setup for end repair, dA tailing, and bar-

coded adapter ligation were handled by the work-

station as were incubations and magnetic bead clean-

ups. Purified, sequenceable libraries (one automated,

one manual) were size selected via agarose gel elec-

trophoresis and excision of a 400 - 500 bp target

range. Amplification was then completed.

Both libraries were assessed to determine quality and

robustness. A Bioanalyzer trace (Agilent Technolo-

gies, Santa Clara, CA) detailed the fragment size dis-

tribution. DNA concentrations were determined with

the use of the PicoGreen reagent (Life Technologies,

Carlsbad, CA). A qPCR quantitation method using a

KAPA Biosystems (Woburn, MA) Library Quant Kit

determined the levels of adapter-bound fragments.

After cluster generation and sequencing of the librar-

ies using one lane on a HiSeq 2000 run (Illumina, San

Diego, CA), the data was characterized.

To ensure the library fragments were appropriately

ligated to Illumina adapter sequences a qPCR library

quantification was completed. The Ct values (Figure

2) and concentration of adapter-bound fragments

(Table 1) for the automated and manual libraries

were similar and of sufficient quality for sequencing.

Both the automated and manually created libraries

amplified equally, passing a common QC metric used

prior to sequencing. To confirm the libraries were a)

representative of the original template and b) of nec-

essary integrity to provide usable raw data, they were

submitted to DNA cluster generation and sequencing on a single lane of a HiSeq 2000 instrument. Sequence Data QC Resulting raw sequence data for 100 bp paired-end

reads were sorted by barcode, converted to FastQ files

and processed with Trimmomatic (Usadel lab, Max

Planck Institute, Potsdam, GER). Adapter sequences

were removed, leading and trailing bases were

scanned, and a sliding window was used to trim reads

at points over which average Q scores dropped below

15.

Figure 1 | Fragment Size Comparison Post Size Selection. Bioanalyzer traces for the automated (a) and manual (b) libraries. A dark electropherogram band in (a) demonstrates the automated process efficiently recovered DNA within the range targeted by the size selection.

Protocol

RESULTS Library QC Bioanalyzer traces for the manual and automated

libraries indicated both had acceptable fragment size

distribution profiles. Both libraries had similar traces;

the manual library had an average fragment of size of

423 bp compared to 417 bp for the automated library

(Figure 1). A PicoGreen experiment indicated total

DNA concentrations of 157.16 nM and 180.66 nM for

the automated and manual libraries respectively

(Table 1).

Table 1 | Determination of Overall dsDNA Concentra-tion (PicoQuant) and Adapter-bound Fragments.

Library PicoQuant (nM)

qPCR (nM)

Automated 157.16 169.2

Manual 180.66 172.5

Differences between the sequence data sets gener-

ated with the two libraries did exist. The automated

set contained approximately 75% more paired-end

reads than the manual set (Table 2). This difference

can be attributed to either 1) highly efficient DNA

cluster generation with the automated library or 2)

enhanced adapter ligation using NEXTflex™ library

preparation technology.

Both libraries appeared to have levels of sequence

duplication that may have been introduced during the

PCR amplification process. FastQC analysis shows that

the automated NEXTflex preparation had a significantly

lower levels of duplication compared to the manual

preparation.

CONCLUSIONS:

In addition to reducing hands on time and the overall

burden of manually preparing hundreds of libraries,

Figure 2 | qPCR Quantification of Automated (a) and Manual (b) Libraries. Log (rRn) vs. cycle count for qPCR analysis of adapter-bound fragments in the automated (a) and manual (b) libraries. A common threshold of 32.7 was used for both, with Ct values of 5.3 and 5.0 in (a) and (b), respectively.

Protocol

An understanding of the sequence characteristics was

necessary to prove the automated method was vi-

able. A FastQC analysis (Andrews lab, Barbraham

Institute, Cambridge, UK) of the R1 read from the

paired-end sequences revealed similarities between

the two libraries for sequence quality and content.

The per base sequence quality and overall read qual-

ity scores (Figure 3) illustrated that both sample

preparations resulted in libraries of high integrity. In

addition, both data sets indicated overall GC content

to be 39% (data not shown). R2 sequence metrics

mirrored those found in R1 (data not shown).

Table 2| Paired-end Reads for Automated and Man-ual Datasets.

Library Reads (M)

Automated 114.1

Manual 65.4

Protocol

automated VERSA liquid handling together with

NEXTflex library preparation has several demonstra-

ble advantages:

High DNA recovery during bead based clean-up

75% greater paired-end reads

Fewer duplicate reads

High read quality

VERSA is a unique liquid handling platform in that it is

an open system, allowing the user to choose their

library preparation technology. With new and im-

proved library technologies, users of closed liquid

handling platforms are at a disadvantage. In addition

to losing flexibility, “closed” system users are de-

pendent on a single manufacturer which can lead to

supply and pricing speed bumps. By using the VERSA

system, users can take advantage of NEXTflex, one of

the most robust DNA library technologies on the mar-

ket. In addition to improved read numbers, coverage

and assembly, NEXTflex libraries have the largest array

of indexed adapter barcodes available.

ACKNOWLEDGEMENTS Sequencing was performed independently by the Uni-

versity of Arizona Genetics Core under the supervision

of Drs. Ryan Sprissler, Matthew Kaplan and Joe Still. Dr.

Michael Nachman provided the samples. Drs. Sikander

Gill and Rajwant Gill provided technical support from

Aurora Biomed’s offices.

REFERENCES 1. Biesecker LG, Mullikin JC, Facio FM, Turner C, Cherukuri PF, Blakesley RW, Bouffard GG, Chines PS, Cruz P, Hansen NF, et al. 2009. The ClinSeq Project: Piloting large-scale genome sequencing for research in genomic medicine. Genome Re- search 19: 1665– 1674. 2. Glenn TC, 2011. Field guide to next generation DNA se- quencers. Molecular Ecology Research 11: 759-769. 3. Farais-Hesson E, Erikson J, Atkins A, Shen P, Davis RW, Scharfe C, and Pourmand N, 2010. Semi-automated library preparation for high-throughput DNA sequencing plat- forms. Journal of Biomedicine and Biotechnology 2010:617469.

-

Figure 3| Sequencing Metrics. Per base and per sequence quality scores for R1 ends of the automated (a, b re-spectively) and manual (c, d respectively) libraries. All median per base scores exceeded Q28, and the average read quality peaked at Q38 for both libraries.