Sensitive and Reliable Variant Detection From Challenging Samples

1. Katja Heitz, Rumeysa Akinci-Tolun, Jennifer Fostel, Anika Joecker and Nan Fang
QIAGEN GmbH, QIAGEN Strasse 1, 40724 Hilden, Germany
110384608/2016
Sensitive and Reliable Variant Detection
From Challenging Samples
Sample to Insight
Abstract
The rapidly developing next-generation sequencing (NGS) technologies provide highly sensitive methods in detecting and
characterizing variants in clinical samples. However, clinical samples are often limited in quantity, as well as compromised
in quality. Such samples are not suitable for standard NGS library construction methods, which commonly require hundreds
of nanograms of good-quality DNA. Examples of such challenging clinical samples include laser capture microdissection
(LCM) samples, formalin-fixed paraffin-embedded (FFPE) samples and circulating DNA.
Here, we describe an optimized workflow to efficiently convert small amounts of DNA samples into sequencing libraries.
The library construction protocol is based on the QIAseq Ultralow Input Library Kit, which has optimized enzyme and
buffer formulations that enable high library conversion rate, as well as unbiased library amplification from as little as
10 pg input DNA. In combination with an optimized hybrid-capture–based target enrichment procedure, the workflow
described in this poster enables reliable variant detection even with low DNA input amount, enabling sequence insights
from challenging sample types.
High mapping rate, conversion rate and
uniformity
100 pg or 1 ng of fragmented genome-in-a-bottle
(GIAB, RM8398; NIST) human reference DNA was
constructed into an NGS sequencing library using
the QIAseq Ultralow Input Library Kit (QIAGEN) and
sequenced on a HiSeq™
4000 (Illumina) to an average
coverage of 23X and 25X, respectively. PicardTools
was used to analyze sequencing data and determine
library quality and complexity. The sequencing libraries
constructed demonstrate almost 100% of reads mapped
to the reference, high conversion rate and even coverage
of genome regions with different GC contents.
QIAseq Ultralow Input Library Kit For Minimal Input
Variant Detection from 1 ng gDNA in WGS and WES
High coverage and uniformity of libraries constructed from FFPE or cell-free DNA samples
A sequencing library constructed using the QIAseq Ultralow Input Library Kit delivers comprehensive and even coverage
when used in combination with hybridization-based target enrichment; essential for sensitive, reliable mutation detection.
Sequencing library construction coupled with hybrid-capture–based target enrichment is commonly used to detect mutations
in clinically relevant samples such as FFPE or cell-free DNA (cfDNA). We tested this workflow using Quantitative Multiplex
Formalin Compromised DNA Standard I or Multiplex I cfDNA Reference Standard Set (both from Horizon Discovery).
FFPE or cfDNA DNA standard (20 ng) was constructed into sequencing libraries and the libraries were enriched for the
127 significantly mutated genes (SMGs) using the xGen Pan-Cancer Panel (IDT) and sequenced using a miSeq®
(Illumina)
sequencer. Data were analyzed using CLC Biomedical Workbench (QIAGEN).
High specificity and sensitivity
The variants in the GIAB samples can be detected with high specificity and sensitivity from 1 ng input DNA using either
whole genome sequencing (WGS) or whole exome sequencing (WES) methods. Fragmented GIAB gDNA (1 ng) was
constructed into an Illumina sequencing library using the QIAseq Ultralow Input Library Kit. The sequencing libraries were
either sequenced directly, or went through the hybrid capture target enrichment procedure, so that only the enriched
exons are sequenced. The high-confidence SNPs in the GIAB samples were detected with high sensitivity (>90%) and high
precision (>99.50%; or <0.50% false-positive rate) in both WGS and WES experiments.
Comprehensive target coverage. Almost 100% of the cancer gene regions
targeted by the xGen Pan-Cancer Panel were detected in the sequencing,
demonstrating highly efficient library construction and hybrid capture even
with limited (20 ng) FFPE or cfDNA samples.
High SNP calling concordance in WGS with as little as 1 ng GIAB gDNA
sample. The genome analysis toolkit (GATK) analysis pipeline was used
for variant calling of the GIAB sample that were sequenced on a HiSeq
4000 (Illumina) to an average coverage of 25X. With 1 ng gDNA input,
94.66% of the characterized high-confidence SNPs in the GIAB sample
were detected with 99.70% precision (0.30% false-positive rate). Excellent coverage uniformity. The target-enriched FFPE and cfDNA libraries
show high target uniformity, with >97% of bases covered at 0.5X mean
coverage or above, and >80% of bases covered at 0.8X mean coverage
or above.
The QIAseq Ultralow Input Library Kit workflow.
QIAseq Ultralow Input Library for Targeted Sequencing
Summary: Optimized QIAseq Ultralow Input Library Protocol
For Sensitive and Reliable Variant Detection
• Novel QIAseq Ultralow Input Library chemistries enable:
Efficient adaptor ligation
High library conversion rate
Superior coverage uniformity regardless of GC content
High-quality sequencing libraries with sub-nanogram input
• For variant detection with WGS and WES:
High sensitivity and specificity from only 1 ng of input DNA
• For targeted sequencing based on hybrid capture:
High sequence uniformity ensures comprehensive and even coverage
Sensitive mutation detection of low frequency variants even with moderate coverage
For up-to-date licensing information and product-specific disclaimers, see the respective QIAGEN kit handbook or user
manual. QIAGEN kit handbooks and user manuals are available at www.qiagen.com or can be requested from QIAGEN
Technical Services or your local distributor.
Trademarks: QIAGEN®
, Sample to Insight®
, QIAamp®
(QIAGEN Group); HiSeq™
, MiSeq®
(Illumina); xGen®
(Integrated DNA Technologies, Inc.). Registered
names, trademarks, etc. used in this document, even when not specifically marked as such, are not to be considered unprotected by law.
© 2016 QIAGEN, all rights reserved. PROM--9937-001
End-polishing
25ºC, 30 min
65ºC, 15 min
Ultralow Input
Ligation
25ºC, 10 min
Library
amplification
Cleanup
Target region covered, %
100
99.81% 99.76%
80
60
40
20
0
FFPE cfDNA
%
100
80
60
40
20
0
Sensitivity Precision
%
100
97.10%
80.60%
97.11%
80.66%
80
60
40
20
0
FFPE cfDNA
Bases covered
at 0.5X mean
coverage or
above
Bases covered
at 0.8X mean
coverage or
above
%
100
80
60
40
20
0
Sensitivity Precision
Sensitive Detection of Low-Frequency Variants
Targeted sequencing workflow for sensitive mutation detection
The combination of QIAseq Ultralow Input library prep and xGen Pan-Cancer Panel enables sensitive mutation detection
down to 1% allelic frequency with even moderate sequencing coverage. With the FFPE reference sample, all characterized
mutations with allelic frequencies from 1–24.5% were accurately detected. With the cfDNA reference sample (harbo-
ring 8 characterized mutations at around 1% of allelic frequency), 7 of the 8 mutations were detected with a moderate
average coverage of ~500 for the target regions. Sequencing data were analyzed using the CLC Biomedical Workbench
(QIAGEN).
Sensitive mutation detection from FFPE reference. All 11 characterized
mutations in the FFPE reference samples were accurately detected with
the targeted sequencing workflow described above.
Sensitive mutation detection from cfDNA reference. With high adaptor
ligation efficiency and target capture uniformity, even the 8 mutations at
~1% allelic frequency were accurately detected with the targeted sequencing
workflow described above, with a moderate average coverage of 500.
Allelic frequency, %
30
25
20
15
10
5
0
BRAF
V600E
T790M
L858R
G13D
G12D
Q61K
E545K
H1047R
D816V
G719S
DE746-
A750
KIT EGFR EGFR EGFR EGFR KRAS KRAS NRAS PIK3CA PIK3CA
Genes and variants
Expected frequency
Detected frequency
Allelic frequency, %
5
4
3
2
1
0
L858R
DE746-
A750
T790M
G12D
Q61K
A59T
E545K
EGFR EGFR EGFR KRAS NRAS NRAS PIK3CA
Genes and variants
Expected frequency
Detected frequency
C
Reads mapped, %
100
80
60
40
20
0
1 ng 100 pg
Input amount
PCT_PF_READS_ALIGNED
PCT_READS_ALIGNED_IN_PARIS
Conversion rate, %
60
50
40
30
20
10
0
1 ng 100 pg
Input amount
Normalized coverage
1.4
1.6
1.8
2.0
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0 10 20 30 40 50 60 70 80 90
GC content
1 ng
100 pg
Even coverage of human genomes with different GC contents.
The sequencing coverage normalized to the mean coverage
(‘Normalized Coverage’) was calculated for genome regions
with GC contents from 20–80%.
High library conversion rate with
human gDNA sample. The same
libraries were further analyzed to
calculate conversion rate. This
was calculated as estimated library
size divided by the total number
of input fragments. 33.16% and
53.94% of the DNA fragments were
successfully converted to sequencing
libraries with 1 ng and 100 pg
input amounts, respectively.
High mapping rate from minimal
human gDNA input amount.
Both 100 pg and 1 ng samples have
99% reads aligned to the reference.
High SNP calling concordance in WES with as little as 1 ng GIAB gDNA
sample. The sequencing library was constructed with 1 ng fragmented
GIAB gDNA using the QIAseq Ultralow Input Library Kit. Following library
construction, exome enrichment was performed with xGen®
Exome Research
Panel v1.0 (IDT). Genome Analysis Toolkit (GATK) analysis pipeline was
used for variant calling of the GIAB WES sample that was sequenced on
NextSeq 500 (Illumina) to an average coverage of 86X. With 1 ng initial
gDNA input, 91.26% of the characterized high confident SNPs in the GIAB
sample were detected with 99.62% precision (0.38% false-positive rate).