The Pharmacogenomics Journal (2003) 3, 77–96
& 2003 Nature Publishing Group All rights reserved 1470-269X/03 $25.00
www.nature.com/tpj
REVIEW
Single nucleotide polymorphism genotyping:
biochemistry, protocol, cost and throughput
X Chen
PF Sullivan
Department of Psychiatry, Virginia Institute for
Psychiatric and Behavioral Genetics, Virginia
Commonwealth University, Richmond, VA, USA
Correspondence:
Dr X Chen, Department of Psychiatry,
Virginia Institute for Psychiatric and
Behavioral Genetics, Virginia
Commonwealth University, 800 E. Leigh
Street, Richmond, VA 23298-0424, USA Tel:
+1 804 828 8124 Fax: +1 804 828 3223
E-mail: [email protected]
ABSTRACT
The large number of single nucleotide polymorphism (SNP) markers available
in the public databases makes studies of association and fine mapping of
disease loci very practical. To provide information for researchers who do not
follow SNP genotyping technologies but need to use them for their research,
we review here recent developments in the fields. We start with a general
description of SNP typing protocols and follow this with a summary of
current methods for each step of the protocol and point out the unique
features and weaknesses of these techniques as well as comparing the cost
and throughput structures of the technologies. Finally, we describe some
popular techniques and the applications that are suitable for these
techniques.
The Pharmacogenomics Journal (2003) 3, 77–96. doi:10.1038/sj.tpj.6500167
Keywords: SNP; genotyping; allele discrimination; cost; throughput
Received: 19 November 2002
Revised: 7 February 2003
Accepted: 10 February 2003
INTRODUCTION
Single nucleotide polymorphisms (SNPs) have emerged as genetic markers of
choice because of their high-density and relatively even distribution in the
human genomes1–3 and have been used by many groups for fine mapping disease
loci and for candidate gene association studies. When a disease gene has been
mapped to a chromosomal region, a high-density SNP mapping or candidate
gene association studies are logical steps to follow. These efforts normally require
a large amount of genotyping. To perform such large-scale SNP genotyping many
SNP markers are required along with high throughput technologies. Efforts
funded by the government and private companies have produced several
millions of nonredundant SNPs2,3 and they are readily available for such studies.
Genotyping technologies have become a significant bottleneck for these
applications despite rapid progress in the field.
Many new technologies for SNP genotyping have been developed in the last
few years. Rapid technological progress makes it difficult to choose appropriate
methods for a given application. In this article we review current technologies,
focusing on the underlying biochemistry and we compare and contrast the
strengths, weaknesses, throughputs and cost structures of these technologies. As
a result of space limitations, we intend to cover only the most popular methods.
More comprehensive reviews were available elsewhere.4–7 We hope to assist
readers in the selection of methods that are most appropriate for their
applications.
A genotyping protocol normally has two parts—biochemical reactions to form
allele-specific products and detection procedures to identify the products. Most
technologies separate these two processes to enhance resolution, accuracy and
throughput although they can be performed in parallel under certain
SNP genotyping
X Chen and PF Sullivan
78
circumstances. In the sections that follow, we will first
outline a general genotyping protocol as a base for our
discussion and then review (i) biochemistries for allelic
discrimination, (ii) mechanisms for product detection and
identification, (iii) cost structures, and (iv) throughputs.
Finally, we will discuss several popular methods in more
details.
GENOTYPING PROTOCOLS
Typically, genotyping protocols start with target amplification and follow with allelic discrimination and product
detection/identification (Figure 1). Depending on the
mechanisms for allele discrimination and the means of
product detection/identification, some or all of these steps
could be combined and processed in parallel. Clearly, the
fewer the steps involved in a procedure, the greater the ease
for operation and automation, which are of critical importance for large-scale operations. However, fewer steps
may not be the best way to attain optimal data quality, cost
and throughput. The key is to balance the gain from
throughput and cost-saving vs the labor and cost of extra
procedures. When the gain outweighs the labor and cost,
more complex procedures can be justified.
Target DNA Amplifications
Human genome has about 3 billion basepairs. The target of
genotyping is usually a few hundreds basepairs. The
challenge of assaying such a tiny fraction of the genome is
a key reason why most genotyping techniques begin with
target amplification. Almost all techniques use the poly-
Figure 1 A workflow chart for SNP genotyping. Typical protocols
consist of three components: target amplification, allelic discrimination and product detection and identification. The component
reactions can be performed sequentially or in parallel. The style of
their execution reflects differences in design philosophy, throughput expectation and cost-saving measurement.
The Pharmacogenomics Journal
merase chain reaction (PCR) for amplification, albeit other
strategies could be used.8,9 In brief, PCR works by using heat
to separate a duplex DNA molecule into two single-stranded
molecules and then copying each of the two single-stranded
molecules with thermostable DNA polymerases and an
oligonucleotide primer. When the DNA polymerases finish
copying the single stranded molecules, each becomes a new
duplex DNA molecule. This process can be repeated 30–40
times, after each round of amplification the amount of
targeted DNA fragment doubles. By the end of the process,
the amount of the targeted DNA fragment has increased 109
times or more. To our knowledge, the Invader technology10
is the only technique that claims to be able to genotype
directly from genomic DNA, although the amount of
genomic DNA required is about 10 times greater than that
for a PCR.
For genotyping purposes target DNA amplification has to
be specific—an amplicon amplifies one and only one locus
in the genome. Primers that amplify multiple loci can lead
to serious genotype errors. Pseudogenes, conserved sequences within gene families and repetitive sequences are
common features of our genome and they can cause
nonspecific amplifications. To avoid this problem known
repeat sequences should be filtered out before designing PCR
primers, and PCR products should be tested before genotyping.
Since most genotyping techniques rely on PCR for target
amplification, its efficiency and cost become a critical issue
for large-scale applications. PCR performs best when it is
used to amplify a single target, with a success rate of 85% or
better. However amplifying one target at a time is not
efficient. It would be ideal to be able to amplify multiple
targets simultaneously. Unfortunately, multiplexing PCR
has serious problems, the success rate drops to 50–70% for
markers and not all targets are amplified equally. Furthermore, the scoring rate (the percentage of samples that can be
genotyped) also goes down dramatically, from better than
95% for single marker PCR to 60–80% for 5–10 multiplexing PCR. PCR is also expensive for large-scale applications (about $0.25–0.35 per reaction). If 10 multiplex can
be performed, the cost could be shared by 10 genotypes. For
clinical applications, PCR license fees are also a significant
factor and this makes non-PCR-based methods appealing.
For all these reasons, PCR becomes the first significant
barrier for large-scale applications.
Following PCR, there is normally a clean-up step to
remove excess dNTPs and PCR primers left over from the
reaction. This clean-up is necessary for all primer extensionbased methods (see below) because the excess dNTPs and
PCR primers interfere with primer extension. The clean-up
can be done enzymatically with shrimp alkaline phosphatase (SAP) and E. coli exonuclease I (Exo I), which inactivate
dNTPs and PCR primers respectively, or physically via gel
filtrations. This step may or may not be necessary for
hybridization- and ligation-based methods. In addition,
other special procedures to prepare single-stranded PCR
products are also performed here in conjunction with the
clean-up.11,12
SNP genotyping
X Chen and PF Sullivan
79
Allele Discrimination Reactions
Allelic discriminations are the core of genotyping. Mechanistically, the discriminating power originates from DNA
polymerases and DNA ligases. The difference in thermodynamics between matched and mismatched DNA duplexes
is also exploited. These allele discriminating reactions
produce allele-specific products with high specificity and
accuracy.
The mechanism used for allele discrimination has substantial impact on the protocol, cost and throughput. From
protocol perspective if a mechanism selected does not
interfere with target amplification, the two can be combined
and executed in parallel. For example, hybridization,
ligation and the 50 nuclease activity of DNA polymerases
do not interfere with PCR process, it is the combination of
these processes with PCR that created the Molecular
Beacons,13 the TaqMan assay14 and the FRET-DOL15 assay.
These single-step, closed-tube assays greatly simplify protocol and minimize user input so that they are readily
amenable to automation.
The mechanism used for allele discrimination largely
determines the specificity and accuracy of the methods.
Typically, DNA ligases have the best specificity, whereas
different activities of DNA polymerases vary in specificity.
The endonuclease activity used for the Invader assay is
highly specific. The DNA synthesis activity, depending on
the way being used, can be very specific (single-base
extension (SBE)) or marginal (allele-specific PCR). The
discriminating power of allele-specific hybridization is
relatively lower. It is important to understand the discrimination power of these mechanisms, it is equally important,
if not more, to understand how a mechanism is devised and
used in a particular method. The overall accuracy and
specificity of a method depend on its amplification strategy,
allele discrimination and detection format. Under optimal
conditions all mechanisms can have acceptable (499%)
accuracy.
could lead to spurious results. All homogeneous methods
use indexes to infer genotypes. Solid-phase-mediated methods can detect allele-specific products directly or indirectly.
The speed of product detection contributes to the
throughput capacity of a method. While the measurement
of fluorescence intensities, absorbance,11 electric charge,28
or mass spectrometry takes only seconds or less per sample,
electrophoresis can take several hours. The preparation
procedures that required before the detection vary greatly
among techniques, and they are important factors when
considering a technique. While some of the procedures are
common to several techniques others are specific to a
particular method. It is often the case that the efficiency of a
protocol makes it stand out.
Most detection systems have built-in mechanisms to
perform repetitive work for many samples, such as 96- and
384-well plates or microarrays. In addition, most systems
dedicated for SNP genotyping have algorithms for automatic
genotype scoring. The protocols and their capacity
for multiplexing, therefore, are the most important factors
in determining the overall efficiency of cost and
throughput.
Detection and Identification of Allele-specific Products
The next step is the detection and identification of allelespecific products. Depending on the method of detection,
an additional clean-up or purification may be necessary.
Most solid-phase-mediated methods require a clean-up
because the substrates of the allele-specific reaction can
produce false-positive signals and complicate genotype
scoring. Examples are mass spectrometry,12,16–21 zip-code
technology,22,23 Illumina color beads,24,25 Orchid SNP-IT
technology11 and all hybridization based methods.26,27
Product detection and identification can be performed in
many different ways. Mass spectrometry detects and identifies the allele-specific products by molecular weight; DNA
sequencing separates products by size and fluorescence
labels. Other methods uses indexes to infer products, such
as fluorescence resonance energy transfer (FRET), fluorescence polarization (FP), luminescence, absorbance and
melting temperature. When indexes are used, it is critical
that the indexes are directly linked to or correlated with the
products, any factor that alters the linkage and correlation
Allele Discrimination based on DNA Polymerases
DNA polymerases are versatile enzymes that have multiple
functions. Their major function is to replicate DNA during
cell division. To ensure the fidelity of genetic information
across generations, DNA polymerases must be highly
accurate and precise. This requires, during DNA synthesis,
the extending end of a DNA strand and the incoming base
match the template exactly. When one of them does not
match, the extension would stop, awaiting other functions
of the polymerase or other enzymes to correct the error. This
DNA synthesis activity has been exploited for genotyping in
the form of SBE, allele-specific extension (ASE) and allelespecific PCR (AS-PCR). DNA polymerases also have 50 and 30
exonuclease activities in order to repair errors and remove
RNA primers used in DNA replication. These activities
form the basis of a mutation detection system (pyrophosphorolysis-activated polymerization29,30) and the TaqMan
assay. The Invader assay was developed from a special
structure specific endonuclease activity of some archael
bacteria.
BIOCHEMISTRY OF ALLELE DISCRIMINATION
Genotyping is a laboratory procedure that identifies the
alleles presented in a given sample. To accomplish this goal,
genotyping biochemistry must be highly specific. A biochemical reaction identifies one and only one allele at a
time. Since multiple reactions can occur simultaneously at
multiple templates and target loci of a sample, collectively
the same biochemical reaction can identify multiple alleles
and multiple loci.
Popular SNP genotyping technologies currently available
are based on one or more properties of these enzymes and
processes: (i) DNA polymerases; (ii) DNA ligases; and (iii)
hybridization.
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
80
Single-base extension
SBE or minisequencing derives from the DNA synthesis
activity directly. DNA polymerases extend a DNA strand
from 50 to 30 and require a template strand to guide the
incoming nucleotides; the extending 30 base must be basepaired with the template. If the 30 extending base is not
base-paired to its template DNA polymerases would not add
new bases to the strand. If the 30 extending base does not
have the 3-hydroxyl group (–OH), the strand would not be
able to extend any further.31 These features form the basis of
SBE.32,33
SBE, literally speaking, is to extend a DNA strand or a
primer by one base. To ensure all primers extend one and
only one base the triphosphate nucleotides used in the
reaction must be a special group, the terminators (ddNTPs),
which do not have the 3-hydroxyl group. For SBE, extension
primers are designed to anneal immediately upstream to the
target polymorphic sites. In the presence of a DNA
polymerase and candidate terminator nucleotides complementary to the polymorphic bases, primer extension occurs
(Figure 2a). The polymorphic bases dictate which terminators are incorporated. The products of SBE are oligonucleotides that are one base longer than their extension primers.
To facilitate detection the terminators can have specific
labels, of which fluorophores are the most popular.6,34–37 By
detecting the fluorescence labels or other moieties11 linked
to the terminators, the genotype of a sample can be inferred.
The major difference between SBE and DNA sequencing is
that SBE identifies only one base (the polymorphic base)
while sequencing can identify up to several hundreds bases
and determine their relative order. For genotyping purpose,
the information obtained by SBE is adequate although the
surrounding sequence is desirable under some circumstances (see below).
SBE produces highly specific products. Many studies have
shown that the error rate of DNA synthesis for commonly
used polymerases is 10-4 or lower. Combined with appropriate detection mechanisms, genotyping methods that use
SBE produce highly accurate genotype data. These methods
include mass spectrometry, microarray-based detection,
FRET and FP detection.
Allele-specific extension and allele-specific PCR
ASE and AS-PCR make use of the difference in extension
efficiency between primers with matched and mismatched
30 bases. DNA polymerases extend primers with a mismatched 30 base at much lower efficiency, from 100 to
10 000 folds less efficient than that with a perfectly matched
30 base. ASE can be performed by itself37 or with a matching
primer to amplify a specific allele in a sample, that is, ASPCR.38,39 ASE requires two specially designed allele-specific
primers that anneal to the target with their 30 bases
matching the two alleles of an SNP (Figure 2b). In AS-PCR,
a common reverse primer is needed to match with the two
allele-specific ASE primers to amplify the target. Both ASE
and AS-PCR produce products as long as their templates;
whereas ASE amplifies the templates lineally, AS-PCR does it
The Pharmacogenomics Journal
exponentially. The allele-specific primers normally bear
labeling tags38,39 to enable their identification. ASE can be
performed in situ on a chip when the two allele-specific
primers are anchored at separate addresses.37 The formation
of extension products implies that the 30 allele-specific bases
match the polymorphic bases in the templates. By detecting
which primer forms the products, a sample’s genotype can
be determined.
Although SBE possesses superior discrimination in comparison to ASE, genotyping results obtained from ASE may
be acceptable.37 As for AS-PCR, the results are largely
dependent on the sequence surrounding the marker and
the nature of polymorphism. It has been widely observed
that while some AS-PCRs can be highly specific, many
others produce ambiguous results. It appears that both the
local sequence and the type of polymorphism affect the
specificity of AS-PCR.40 Furthermore, the nature of exponential amplification of AS-PCR makes the decay of
discriminating power much quicker than that of linear
amplification as seen in both SBE and ASE. For example, if
the polymerase makes 1% error in at the polymorphic base
or a discriminating factor of 100 to 1 for a homozygous
sample, the accumulated error in the amplified population
would reach B24% for AS-PCR (1–0.9930) after 30 cycle
amplification, but that for SBE and ASE remains at 1%.
When the amplified population is assayed, if only 1% of the
population contains error, it normally goes undetected, but
if that fraction increases to 24%, it surely will be detected
and the sample will be scored as heterozygous. Therefore, it
actually requires higher discriminating power for AS-PCR to
arrive at same error ratio in the amplified population.
Synchronized DNA synthesis—pyrosequencing
Pyrosequencing is a novel approach for DNA sequencing,41,42 which takes advantage of a by-product of the
DNA polymerase reaction when it incorporates a nucleotide
onto a primer. When a DNA polymerase incorporates a
nucleotide onto the 30 end of a primer, it produces a
pyrophosphate. On pyrosequencing, this pyrophosphate
then triggers an enzymatic cascade (ATP sulfurylase, luciferase) to produce a detectable luminescence signal (Figure 2c).
For this scheme to work, the synthesis of DNA strands must
be synchronized, which can be accomplished by adding
nucleotides to the reaction one at a time. The synchronization ensures that no reaction occurs when the incoming
nucleotide does not match the template. When the incoming nucleotide matches the template, the DNA strand
extends one base and a pyrophosphate is released. The
pyrophosphate then initiates the enzyme cascade to produce light; if not, the nucleotide is degraded by yet another
enzyme, an apyrase, to refresh the system. Repetition of this
procedure produces a sequence complementary to that of
the template. Like DNA sequencing, pyrosequencing provides sequence context surrounding polymorphic sites.
Pyrosequencer detects the luminescence in real time and it
can sequence a short stretch of DNA (10–15 bases) in about
20 min.43
SNP genotyping
X Chen and PF Sullivan
81
Figure 2 The biochemistry of allelic discrimination. For all reactions, substrates are listed on the left side and products on the right. (a)
SBE. When an extension primer anneals to the template (underlined) immediately upstream of the target polymorphic site (bold) DNA
polymerases incorporate ddNTPs onto the 30 end of the primer. The bases incorporated are dictated by the polymorphic bases. The
products of SBE are one base longer than the extension primer with an allele-specific label. (b) ASE and AS-PCR. ASE and AS-PCR use allelespecific primers, one for each allele. The 30 end of ASE primers anneal onto the polymorphic bases. If the last base matches the polymorphic
base, the primer gets/is extended, otherwise there is no extension. The products of the extension are normally the same length as the
templates. To facilitate the identification of the products the primers bear distinct labels. (c) Pyrosequencing. Primers for pyroseqeuncing
anneal to templates a few bases upstream from the target polymorphic site. Deoxyribonucleotides are added to the reaction one by one. If
the base added matches the template, the primer extends one base, and produces a pyrophosphate, which is turned into a
chemiluminescence signal via an enzyme cascade. Otherwise the base is degraded by an apyrase provided in the system. (d) Structurespecific cleavage. Three probes, one invader probe (below the template) and two allele-specific probes (above the template), are used for
the Invader assay. One allele-specific probe and the invader probe are needed to form a bifurcated structure with the template. The 30 base
of the invader probe does not have to match the template. If the allele-specific probe matches the template at the target site, a bifurcated
structure with one base overlap between the allele-specific probe and the invader probe would be formed. The one base overlap is
necessary and sufficient to cause the cleavage of the allele-specific probe. The products of the reaction are the 50 portion of the allelespecific probes. (e) Ligation. Ligation joins two probes together. When the bases on both sides of the ligation site match the template,
ligation products are generated, otherwise there are no products. (f) Hybridization. When an allele-specific probe anneals to the target
perfectly it is more stable. If there is a mismatch with the template it would be less stable and can be washed away. The results of
hybridization are that the matched probe remains hybridized and the mismatched is not.
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
82
Like SBE, pyrosequencing uses the synthesis activity of
DNA polymerases and it can generate highly accurate
genotypes. Since dNTPs are introduced in the system one
by one, each addition of a base to the primer produces an
amount of pyrophosphate directly proportional to the
amount of templates, which dictate what and how many
bases are incorporated onto the primer. The signal producing steps (ATP sulfurylase and luciferase) are also quantitative. All these features render the method quite
quantifiable when the template sequence does not contain
a homo string of same bases. When it encounters a homo
string of 4–5 bases such as AAAA or TTTTT, its signal
becomes disproportional and the quantification becomes
unreliable.
Structure specific cleavage — the Invader assay
Some DNA polymerases from archael bacteria have evolved
with an endonuclease activity that requires specific substrate
structure. The structure is a bifurcated duplex with a free 50
end.44 The key feature of the structure is that there must be
at least one base overlapping between the strand that has a
free 50 end and the strand that is annealed to the template.
The enzyme cleaves the strand with a free 50 end when one
or more overlapping bases are detected (Figure 2d). The
cleavage site is dependent on the length of overlapping
bases, eventually it would cut out all overlapping bases to
form a structure that can be readily joined by DNA ligases
(so it is believed to be part of the function to repair DNA
damage45). There would be no cut if there is no overlap. The
Invader assay design uses three oligonucleotide probes, two
allele-specific ones and one common (the invasive probe), to
create artificial bifurcated structures with the templates at
the polymorphic site.10,46 When the appropriate target is
presented, one or two of the allele-specific probes would be
cleaved. The released 50 end of the allele-specific probes can
then form still another bifurcated structure designed with an
arbitrary sequence purely to amplify the signal for better
detection sensitivity (Invader square).
The Invader assay is highly specific and can genotype SNP
directly from genomic DNA. It is a useful alternative to the
methods that based on PCR amplification. For the Invader
assay, the quality of the invasive probe is critical. As stated
above, the cleavage site depends on where the 30 of invader
probe ends. The 30 of the invader probe has to be on the
polymorphic site, one base offset can result in false-positive
and false-negative signals.
The 50 exonuclease activity–the Taqman assay
The TaqMan assay exploited the 50 nuclease activity of DNA
polymerases. During DNA synthesis if the extending strand
encounters a small piece of DNA that is perfectly matched
with the template the polymerase cleaves it. In normal DNA
synthesis, this activity is believed to be responsible for
removing Okazaki fragments. If the small piece of DNA is
not annealed to the template with sufficient stability (ie
there are mismatches between the strands), the small piece
would be pushed off the template but not cleaved.14 The
The Pharmacogenomics Journal
TaqMan assay exploits this feature and designs two dually
labeled fluorescent probes annealed at the SNP site. Each of
the two probes anneals to one allele to create a stable
structure that would lead to its degradation by the moving
DNA polymerase, but when it anneals to the other allele, it
forms a structure that is less stable so that the probe gets
pushed off the template without being cleaved. The cleavage
of the dually labeled probes changes the status of FRET
between the two fluorophores, providing a mechanism for
its detection (FRET detection, see below).
The TaqMan assay also exploited the difference in stability
between the perfectly matched and single-base mismatched
duplexes. It is an example where multiple mechanisms for
allele discrimination are used. Since the difference between
the perfect match and single-base mismatch is subtle, the
TaqMan assay normally requires substantial optimization to
make it work. Although the use of minor groove binding
agents improves the performance of the assay,47,48 probe
design remains largely empirical that poses substantial
difficulties for new users.
Allele Discrimination based on DNA ligases
DNA ligases join the ends of two oligonucleotides annealed
next to each other on a template. The ligation reaction
occurs only if the oligonucleotides are perfectly matched
with their templates at the ligation site. Mismatches at
either end of the ligation site would result in no ligation
products (Figure 2e). Taking advantage of this property, two
oligonucleotides can be designed to anneal to both sides of
an SNP. By detecting the formation of ligation products,
genotypes of the target SNP can be inferred.15 To use DNA
ligases to discriminate the two alleles in an SNP requires
three oligonucleotides, one for each allele and one common
to both alleles. Although DNA ligase can discriminate
mismatches on both sides of the ligation site, it is more
sensitive to the mismatches on the 30 end,49–51 so most
designs use two allele-specific oligonucleotides on the 30 end
and one common oligonucleotide on the 50 end. There was a
recent report that DNA ligase could be used similarly for
typing variations in RNA molecules.52 Ligation is a different
biochemical reaction than PCR, this makes it possible to
combine the two reactions together and to execute them in
parallel.
There are some recent developments that take use of DNA
ligase to create allele-specific, circular-single-stranded
probes (padlock probes), which are then amplified by rolling
circle amplification.52–55 Although these methods are not
yet available commercially, they have potentials to be
developed into homogeneous and isothermal technologies
that can overcome the barrier of PCR amplification.
Allele Discrimination based on Hybridization
DNA hybridization is based on the intrinsic property of DNA
molecules that they form duplex structures if complementary strands are present. Under many conditions, the two
strands forming a duplex structure do not have to be 100%
complementary. This is very important as more mismatches
between the two strands make the duplex structure less
SNP genotyping
X Chen and PF Sullivan
83
stable thermodynamically. A less stable duplex is easier to
disrupt (denature) when it is heated. The temperature that
disrupts 50% of duplex DNA molecules is defined as melting
temperature (Tm) for that DNA molecule. SNP genotyping
by hybridization exploits the subtle difference in thermodynamics between a perfectly matched and a single-base
mismatched duplexes. This method designs two allelespecific probes to hybridize to the target; each probe forms
a perfect duplex with one allele but forms a mismatched
duplex with the other allele. The ways to differentiate the
two duplexes form the basis of detection. This can be
accomplished via steady-state or dynamic hybridization.
DNA hybridization—steady-state measurement
For SNP genotyping, DNA duplexes formed by the two
alleles differ only by one base pair. The difference in
thermodynamics between the two duplexes is subtle in
most cases and its detection requires sophisticated design
and instrumentation. One way to overcome this weakness is
redundant design, which uses multiple probes for each
allele.27,56 A consensus signal from those probes designed
for the same allele is used to score the genotypes. The locked
nucleic acid (LNA), a new analog of nucleotide, has been
reported to increase the Tm of hybridization probes and to
help discriminating matched and mismatched probes.57,58
For steady-state measurement, DNA hybridization is
normally performed under optimal conditions that probes
anneal only to their corresponding allele but not to the
other allele. When the reaction reaches steady state,
unhybridized probes are washed away, Figure 2f. Genotypes
are scored based on the probes that remain hybridized. In
traditional hybridization, targets are anchored on solid
support and probes are labeled and put in solution. In order
to assay a large number of SNPs with microarrays the process
is now reversed, probes are anchored on specific addresses of
a microarray and targets are labeled and put in solution.
Labeling of target DNA is commonly done in conjunction
with PCR amplification by using fluorescence-labeled
dNTPs. Genotypes are scored by detecting fluorescent
intensities at specific addresses that hold allele-specific
probes for each SNP.26,27
probes cheaper. It relies on the use of duplex-dependent
fluorescence, a property of some fluorescent dyes that emit
only if they are intercalated into a DNA duplex. As
temperature is gradually ramped up to denature the
duplexes a steep decrease of fluorescent intensity would be
observed when temperature reaches the Tm, indicating that
the duplexes are melting into single-stranded form. Dynamic measurement is a real-time process that demands
sophisticated temperature control coupled with data acquisition. This limits its utility for some applications.
The allele-specific reactions described above can be
performed in aqueous solution32,38,39,60–63 or anchored to
a solid support.36,37 The choice of a particular reaction
format depends on the characteristics of the technology and
throughput expectations.
MECHANISMS FOR PRODUCT DETECTION AND IDENTIFICATION
Detection mechanisms for allele-specific products vary
greatly, from fluorescence and luminescence reading to
mass and electric charge measurement. The mechanisms
roughly fall into two categories: homogeneous detection
and solid-phase-mediated detection. Homogeneous detection is more amenable to automation as it does not require
purification or separation after the allele discrimination
reactions. However, because the products are not separated
or purified, homogeneous methods have limited capacity
for multiplexing. Another feature of all homogeneous
methods is that they are collective measurements that
depend on the relative amounts of reaction substrates and
products. These methods detect the products only if the
products become dominant. If the substrates remain
dominant, these methods would not be sensitive enough
to identify the products even if they are formed in the
reactions. In other words these methods are more prone to
false negatives. In contrast, solid-phase-mediated methods—
because of their separation and purification steps—offer
superior detection sensitivity and greater multiplexing
capacity. Detection formats have significant impact on
throughput and cost and are one of the important factors
when considering SNP genotyping technologies.
Dynamic allele-specific hybridization—DASH techniques
In contrast to the end point measurement of steady state,
dynamic hybridization measures denature kinetics of DNA
duplexes formed by allele-specific probes and their targets.59
Since duplexes formed by the two alleles have different
thermodynamics, they would have different Tm. In the
process, the duplexes are heated gradually over a range of
temperatures that cover the Tm of both alleles while
multiple data points of fluorescence intensity are collected.
The data create melting curves that are characteristic of each
allele. Tm profiling characterizes the hybridization behavior
of allele-specific probes better than a single-point measurement.
The measurement of Tm does not require modification or
fluorescence labeling of allele-specific probes, this makes the
Homogenous Detection Mechanisms
Fluorescence resonance energy transfer
In recent years, there has been a great increase in the
applications of fluorescence in the field of DNA testing or
genotyping, of which FRET is one of the most popular
approaches.62 FRET is a physical phenomenon that occurs
between two fluorescent groups when they are in physical
proximity and one fluorophore’s emission spectrum (the
donor) overlaps the other’s (the acceptor) excitation spectrum. When FRET occurs the donor emission is quenched
and the acceptor emission increases when the donor
is excited (stoke shift), Figure 3a. Both the quenching of
donor emission13,14,64,65 and the increasing of acceptor
emission66–68 can be used to monitor FRET.
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
84
For SNP genotyping, there are two different strategies that
FRET can be used for detection. The first is based on the
separation of two closely spaced fluorescent groups.13,14,38,69
The separation interrupts the energy transfer or greatly
reduces its efficiency. The second is to stitch two fluorescent
groups together from distant spaces to promote energy
transfer between the two fluorophores.15,70 To apply FRET
The Pharmacogenomics Journal
principle for SNP typing, the reactions can be devised so that
the formation of allele-specific products is correlated to the
occurrence or disappearance of FRET phenomenon. Once a
direct correlation is established between the reactions and
FRET, monitoring of FRET during the allele discrimination
reactions can be used to infer the formation of allele-specific
products and thus to identify the genotypes of a sample.
SNP genotyping
X Chen and PF Sullivan
85
FRET has been successfully used to detect allele-specific
products from SBE,63 ASE,38,71 DNA hybridization,13,14,72
DNA ligation15 and structure specific cleavage.10 It is the
most popular mechanism for homogeneous detection.
Several commercially available SNP genotyping technologies, such as the TaqMan 50 nuclease assay (http://www.appliedbiosystems.com), the Molecular Beacons (http://
www.molecular-beacons.org/), the Invader assay (http://
www.twt.com), the scorpion AS-PCR assay (http://
www.dxsgenotyping.com/technology_main.htm), and the
AS-PCR assay (http://www.intergenco.com/snp.html) all
use FRET as the detection method.
Fluorescent polarization
FP is another property that has been widely exploited for
biological applications, although its application for genotyping was relatively recent event.35,73 When a fluorophore
is excited by plane-polarized light, its emission remains
polarized if the molecule is still and the angle between the
exciting plane and the emitting plane is a function of the
mass of the molecule when other parameters are kept
constant. In reality, emissions are depolarized even if the
molecules are excited with polarized light because the
fluorophore molecules are in constant motion and molecular motion cannot be synchronized. What observed is a
collective measurement of a population of molecules. The
degree of depolarization is inversely proportional to the
3
Figure 3 Product detection by fluorescence. (a) FRET. FRET occurs
when two fluorophores are in close proximity and one fluorophore’s (the donor) emission overlaps the other’s (the acceptor)
excitation spectrum. When FRET occurs the donor’s emission
decreases (quenched) whereas the acceptor’s emission increased
(stock shift). FRET decays dramatically when the distance between
the fluorophores increases, because it is inversely proportional to
the sixth power (1/106) to the distance. Both the quenching of
donor emission and the increasing of acceptor emission can be used
to monitor FRET. There are many designs to use FRET for product
detection, of which the TaqMan 50 nuclease assay, shown here, is
one of them. Others are described in the text. (b) FP. FP is
proportional to molecular volume when pressure, viscosity and
other parameters are constant. If the size or molecular volume of
the fluorophore changes the FP property of the fluorophore will
change. In the SBE reaction shown here, the molecular volume of
the dye terminators increases when the terminators are incorporated onto primers, this will lead to an increase of FP values of the
fluorophores linked to the terminators. (c) Fluorescence intensity
detection. Fluorescence intensities of allele-specific products can be
measured directly if the products are separated or purified.
Microarrays are one of the many commonly used media to sort
and separate the products. A microarray, as shown, contains
thousands or more addresses, each of them is anchored with a
predefined DNA sequence. For each SNP, a sequence complementary to one of the predefined sequences is used to tag the SBE
primer. SBE reactions for many SNPs can be pooled and products
are hybridized to the microarray. Different SNPs are sorted by these
anchored sequences to unique addresses. The measurement of
fluorescence intensities at an address identifies genotypes for an
SNP.
mass of these molecules. Applying this principle to allele
discrimination reactions, FP values (measurements of the
degree of polarization) change as the reactions proceed as
the substrates are turned into products. In the case of SBE,
for example, the FP value of the fluorescent dye linked to a
terminator increases (becomes less depolarized) as more dye
molecules are converted from nucleotides into larger
oligonucleotide molecules (20–30 times bigger for commonly used primers consisting of 20–30 bases), Figure 3b.
Thus, the measurement of the FP value serves as an index for
the allele discrimination reactions, which depend on the
genotypes of DNA samples.
Many factors can influence FP measurement, including
temperature, viscosity and nonspecific interactions between
the fluorescent dye and other buffer components or surfaces
of the reaction vessel. So, it is important to maintain
constant environments between experiments and samples.
Deviation from these constants can lead to false-positive or
false-negative results. The FP principle has been successfully
applied to SBE,35 DNA hybridization,58,73 the Invader assay
61
and the TaqMan 50 -nuclease assay.74
Solid-phase-mediated Detection Methods
The use of solid support in the detection step could
potentially increase throughput and reduce cost significantly because it enables massive parallel detection and
analysis. A common use of solid supporter is to provide a
vehicle to sort and purify allele-specific products so that the
detection of these products can be more sensitive and
efficient. DNA microarrays and fluorescence coded microbeads are commonly used supporters and DNA hybridization is
the mechanism to sort allele-specific products. When a
product is purified and/or sorted to a supporter or an
address, the measurement of fluorescence intensities and
colors at the supporter or array address can be used for the
identification of the product.
Identification of products by mass spectrometry
Mass spectrometry measures precise molecular weight of a
molecule. Since allelic discrimination reactions always
change the weight of a molecule, mass spectrometry can
be used to identify products from a reaction.12,16–21,75
Among the four bases that form DNA molecules, the
smallest difference is between the A and the T bases (about
9 Da) and discriminating an A/T polymorphism can be
difficult. A modification of SBE using tagged ddNTPs could
make the difference easier to detect.76 Another approach is
to use a combination of ddNTPs and dNTPs for SBE reaction.
The combination of the ddNTPs and dNTPs is made such
that one allele stops at the polymorphic site, the other allele
stops at one or few bases downstream. This makes the
products different in size and, therefore, much easier to
differentiate.77 MS detection is capable of multiplex. A
simple way to multiplex is to use tagged extension primers.
Extension primers with different tags shift their masses to
different ranges, thus multiple SNPs can be detected
simultaneously.21 Higher throughput is the selling point
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
86
for mass spectrometry. To make the approach viable 3–5 multiplexing is expected otherwise its lengthy protocol
would be inefficient.
It is worth pointing out that MALDI-TOF MS is one of the
best quantification methods for allele frequency estimation.
As the throughput of individual sample genotyping imposes
a serious limit for many studies, more and more people are
looking for ways to reduce the number of genotyping.
Genotyping pooled samples becomes attractive. For that
purpose, evaluation of allele frequency estimation for
different technologies becomes necessary. Multiple recent
studies conclude that SBE combined with MS detection
achieves highly accurate frequency estimates for pooled
samples, the difference between the pooled and individually
genotyped frequencies is 1–2%.78–82
Microarray as supporter for genotyping
The most important feature of microarray is its high density.
In a small area (1–1.5 cm2) hundreds of thousands of
predefined sequences can be arrayed. These predefined
sequences sort products from allele-specific reactions via
DNA hybridization. When allele-specific reactions are
coupled with fluorescence labeling, either on the primers
or on the nucleotides, simple measurement of fluorescence
intensities at addresses of a microarray identifies the
products from the allele-specific reactions.
Products from SBE, DNA ligation and structure-specific
reactions can be engineered to use microarray supporter. A
common practice is to attach a short sequence to the
primers involved in the allele-specific reactions. The
attached sequence is complementary to one of the sequences predefined in the array. For each marker there is a
unique sequence. These marker-specific sequences or tags
sort products from the allele-specific reactions to markerspecific addresses on the array. A leader in DNA chip
manufacture, Affymetrix Inc. (http://www.affymetrix.com/
index.affx), has developed a universal DNA chip (GenFlext
Tag Array) that features 2000 unique DNA sequences. To use
this universal DNA chip one can, in principle, score 2000
SNPs with one hybridization given that one can do 2000
allele-specific reactions efficiently and effectively.23
Another way to use a microarray for genotyping is to array
the entire target sequences on the chip using multiple
overlapping oligonucleotides. Several large-scale SNP genotyping reports used this strategy. In essence, this approach
resequences the target by hybridization.26,27,83 When the
entire sequence of the target is resequenced, genotypes
become clear. An advantage of this approach is that novel
SNPs can be discovered and genotyped in the same process.
The major limitation of this approach is its cost—all reports
that use this approach came from high profile groups with
industry sponsorship. The resequencing approach also
suffers from low accuracy (80–90%) and it is normally not
appropriate for linkage disequilibrium and association
studies.
The microarray approach is capable of high throughput,
but it is limited to those applications that use relatively
The Pharmacogenomics Journal
small numbers of subjects and large numbers of SNPs. When
a large number of subjects is used it is difficult to
accommodate multiple subjects on a single array. Another
practical barrier for using microarray for genotyping is how
to generate efficiently a large number of allele-specific
products before hybridizing to the microarray.
Fluorescence coded microbeads as supporters
Similar to DNA chips, well-characterized and unique DNA
sequences could be attached to microbeads that feature
unique fluorescence signatures.22,24,25,84 After allele-specific
reactions products are hybridized to the microbeads. The
hybridization links the products of an SNP to microbeads
with a unique signature. The identification of the microbeads by their unique signatures could lead to the identification of allele-specific products hybridized to them,
therefore, the identification of genotypes for the SNPs.
Like microarray, color-coded microbeads can process a
large number of SNP markers in parallel. In addition to that
microbeads offer more flexibility that is not easily implemented in DNA chips. This is because each microbead is
independent of other microbeads and can be manufactured
and used separately. When 10 or 100 SNPs are processed, 10
or 100 types of microbeads are mixed with the products
while 10 or 100 SNPs only use a small fraction of the features
encoded in a DNA chip. The flexibility of color-coded
microbeads offers more competitive cost than that of DNA
chips.
Separation of products by electrophoresis
Electrophoresis has long been used to separate and identify
DNA molecules in applications like DNA sequencing. In
electrophoresis, DNA molecules are separated by size. For
allele discrimination, the formation of allele-specific products always results in changes in probe sizes. When the
products are labeled with fluorescence tags they can be
readily separated and identified by electrophoresis.83,85
Tedious protocols and relatively high costs are main
obstacles against the use of electrophoresis for genotyping.
The implementation of capillary electrophoresis has reduced
the labor required. However, until more SNPs can be packed
into one lane/capillary the cost will remain a substantial
barrier for broad use of electrophoresis for SNP genotyping.
There are some commercial products, including the SNaPshot from Applied BioSystems and the SNuPe from Amershan Biosciences (http://www.amershambiosciences.com),
which are developed primarily for SNP genotyping by using
DNA sequencers. These products are convenient for those
users who have access to sequencers.
COST STRUCTURE OF SNP GENOTYPING
The cost of genotyping has roughly three general components in addition to technician’s time: (i) Instrument and
initial setup costs; (ii) fixed cost per marker and (iii)
consumable costs proportional to the amount of usage.
Technician time depends on the protocols. Protocols that
SNP genotyping
X Chen and PF Sullivan
87
are amenable to automation could potentially save greatly
on labor cost, and most homogenous techniques have this
potential.
Instrumentation and initial setup costs are normally high
for most technologies and the range is quite wide, somewhere from $25 000 to over $1 000 000. Since many
techniques use fluorescence to label allele-specific products,
they need instruments for fluorescence detection. This is by
far the largest group of instruments used for SNP genotyping, including fluorescence microplate readers, microarray
scanners, DNA sequencers and real-time fluorescence detection systems. The selection of instrument normally depends
on the nature of the technology selected, which in turn
depends on the specific application and throughput expectations. Dedicated systems usually are easier to learn and
to use, and have better technical support and higher
throughput. They may also possess automation features
and include scoring algorithms/software. In contrast, a
versatile system could potentially be shared by multiple
groups and applications, but for each application the users
have to spend substantial time and effort to make it work.
The high cost in instrument investment makes it a
significant factor when considering a new technique.
Fixed costs for each SNP are specific to each technology.
These include the number of oligonucleotides used for
each SNP and special modification and handling of
these oligonucleotides. For the number of oligonucleotides
used, AS-PCR and primer extension methods are
most efficient, using three oligonucleotides. AS-PCR
uses all three oligonucleotides for PCR, while
primer extension methods use two for PCR and one for
extension. Hybridization methods normally use four oligonucleotides, one for each allele, and two for PCR. Some
designs use more oligonucleotides for each allele to increase
specificity and signal/noise ratio. Ligation based methods
use three oligonucleotides in addition to two for PCR.
Finally, the Invader technology uses 5–7 oligonucleotides
depending on whether PCR is used or not. Of these
oligonucleotides, three are specific for a given marker and
are responsible for generating primary signal. Two oligonucleotides, one for each allele, are used for secondary signal
amplification and these two oligonucleotides are universal
for all markers. If PCR is used, two additional primers are
needed.
The more the oligonucleotides used for an SNP, the higher
the fixed cost. However, if the number of samples to be
genotyped for an SNP is sufficiently large, fixed cost for
oligonucleotides becomes negligible. Another fixed cost
item is the use of specialty reagents and processes, including
special purification and/or modification of oligonucleotides.
For example, mass spectrometry requires HPLC purification
for extension primers; because of its high sensitivity, 1
and 2 oligonucleotides resulted from standard oligonucleotide synthesis could cause problems. Pyrosequencing
uses a biotin-labeled primer to facilitate template preparation. The synthesis and purification of biotinylated
oligonucleotides are much more expensive than
regular oligonucleotide. The TaqMan 50 nuclease assay
and the Molecular Beacons use two doubly-labeled
fluorescence probes that are quite expensive ($200–400/
probe, compared to $20–25/probe for regular oligonucleotides). Many other methods use fluorescence labeling for
oligonucleotides or terminators. Specialty reagents are also
used to modify the oligonucleotide backbone to facilitate
purification and separation of either PCR or extension
products.12 These specialty reagents normally are specific
to one or a few methods, and can increase cost remarkably.
The requirement for specialty reagents is a potential
weakness.
Consumable costs cover enzymes and reagents used for
the reactions, plastics for sample transfer and clean-up and
microarrays and microbeads for sorting and purification,
etc. Some of these items are specific to a method while
others are common to many. For examples, the Invader
technology and pyrosequencing use special enzymes for
their allelic discrimination and detection reactions, but
almost all SBE-based methods use dye-labeled ddNTPs and a
special Taq DNA polymerase86 that incorporates the ddNTPs
much more efficiently.
Most vendors compare costs based on consumable costs
proportional to the volume of use. However, such limited
comparisons are inappropriate, and often misleading because consumables do not include fixed costs, the complexity of protocol, the hands-on time of technician and initial
instrument investment. A fair comparison, we believe,
would be the number of genotypes accomplished by an
employee with a fixed budget and time. Applications
and throughput structures should also be taken into
account (see below). When these apply to users, they may
have different options and need to consider existing
instruments, the expertise and experience of their employees and find out if that are compatible with the new
technique. As lack of expertise is a common cause seen for
the failure to implement new technologies, appropriate
training of employees should also be included in the
overall cost.
Understanding the protocols and cost structures of the
technologies provides a base to compare and contrast the
capacities and costs of different technologies, and to
determine which technology might be most suitable for a
given application. Generally speaking, prices for the technologies currently available on the market vary significantly,
between $0.50–2.00/genotype for consumable reagents. At
the low end of the price range, the template-directed dyeterminator incorporation assay with fluorescence detection
(FP-TDI, see below) is roughly $0.50/genotype for reagents
by list price. Some other methods, because of their
requirement for special reagents and/or clean-up procedures, the price is over $2.00/genotype (pyrosequencing).
MALDI-TOF mass spectrometry is about $0.75 without
multiplexing. For the array-based systems, Affymetrix’s
GenFlex costs about $500–650/chip, microbeads from
Illumina (http://www.illumina.com/prod_genotyping.htm)
need specialized instrument to read (the system costs more
than $1 million), its contract service costs $0.15–0.2/
genotype with a starting job of $100 000.
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
88
TECHNOLOGY, APPLICATION AND THROUGHPUT
STRUCTURE
Each technology, because of its protocol design and
detection format, has certain constraints on throughput.
From the throughput point of view, PCR amplification is a
critical and limiting step. When PCR is used to amplify a
single SNP, over 85% of SNPs will be successfully amplified
and for each SNP 95% or more samples will be scored for
genotypes. This seems acceptable for most applications.
However, amplifying one SNP at a time would significantly
limit the throughput and increase the cost. Some technologies hope to increase their throughput by multiplexing
PCR. But there are problems. When multiple SNPs are
amplified together, only 50–70% of the SNPs can be
amplified successfully. Even for those SNPs that are amplified, the amount of products varies greatly, somewhere
between 10 and 1000 folds. What makes the scenario even
worse is that the calling rate for samples decreases
dramatically—some SNPs that are scored for sample A may
not be scored for sample B or C. As a result, many samples
will not have genotype data. For applications that could not
tolerate missing data, the strategy to use multiplex PCR to
increase throughput would not work. The limitation
imposed by PCR capacity applies to almost all technologies
except the Invader technology. Even for the Invader
technology, limited PCRs are still necessary for many SNPs.
However the impacts of PCR limitation vary among
different technologies. For most homogenous detection
formats like FRET and FP, because their detections have very
limited capacity of multiplex, they may not need to rely on
multiplexing PCR to increase their throughput. Constraints
imposed by multiplexing, therefore, are not important for
these techniques. For other formats like mass spectrometry,
electrophoresis and microarray hybridization, the capacity
of PCR multiplex would greatly increase their throughput.
In other words, these techniques are much more dependent
on PCR multiplexing than others to reach their throughput
potential.
The cost structure has significant impact on throughput.
For technologies with high fixed cost/SNP, the efficient way
to increase throughput is to increase sample size such that
the fixed cost can be shared by many samples. This
requirement would limit the applications to which these
technologies can apply. The TaqMan 50 nuclease assay, the
Molecular Beacons and the Invader assay belong to this
group. Microarrays have different cost structures that are
conditioned on the effective usage of the features in the
arrays. To reach the potential throughput, all addresses of an
array should be used, otherwise it would be inefficient or too
expensive. This approach requires relatively large and fixed
number of SNPs because the physical addresses available of
the arrays constrain flexibility of the number of SNPs.
Different applications have different throughput structures. For example, in the clinical/diagnostic setting,
relatively few SNP markers are to be genotyped, but the
sample size is potentially very large. Under these conditions,
the fixed cost and optimization for markers becomes trivial,
because these costs could potentially be shared by many
The Pharmacogenomics Journal
samples. Methods like the TaqMan 50 nuclease assay,87–89 the
Molecular Beacons90 and the Invader technology91 that
have relatively high fixed cost and/or require extensive
optimization fit the clinical/diagnostic setting very well.
Other applications use large numbers of SNPs but relatively
few samples. Here a microarray approach may be more
reasonable.27,92 Finally, there are applications that require a
relatively large number of both SNPs and samples (eg, both
in the thousands or more). Examples include gene mapping
studies by linkage or linkage disequilibrium for complex
diseases and traits93 and pharmacogenetics.94. Neither of the
above approaches is suited to these applications. This group
of applications demands the most cost effective, high
throughput and flexible technologies because for these
applications the SNPs used, the number of SNPs and sample
sizes vary greatly from study to study.
Owing to the differences in requirements, the limitation
of throughput is not exactly the same thing across applications. For some of these applications, a throughput of
thousands of genotypes (samples markers) per day would
be sufficient, and most techniques described in the section
on popular methods are capable of delivering this throughput. Projects with genotyping needs from 104–105 genotypes
would be realistic and could proceed with these technologies. Such applications include small association studies and
fine mapping of chromosome loci for disease-susceptible
genes. Projects that require 107 or more genotypes (ie whole
genome linkage disequilibrium/association studies) would
not be feasible without significant technological advances.
CRITERIA FOR SELECTION OF TECHNOLOGIES
With good understanding of protocols, allele discriminations, detection formats, cost and throughput structures, we
would now like to discuss how to evaluate and select
technologies that are appropriate for given applications. It
should be kept in mind that high throughput SNP typing is a
rapidly changing field and many competitive and compelling methods are under development and can be developed
into mature and commercial products. The best technologies today are likely to become outdated in near future.
General criteria for selecting a technology are accuracy/
sensitivity, success rates, flexibility, cost and throughput.
Except for clinical applications, accuracy and sensitivity
seldom are an issue, as the accuracy and sensitivity of all
commercially available techniques have been tested and
verified by many independent groups. For example, there
are many reports for SBE with MS detection,12,16,17,20,76,95
SBE with microarray23 and microbeads96–99 detection, the
FP-TDI,100–102 the TaqMan assay,64,88,103 pyrosequencing,104–106 the Invader assay46,107–109 and ligation-based
methods.53,84,110,111 The issue becomes more salient when
users intend to try out newly developed but not widely used
methods. In that context, parallel and systematic tests with
established methods should be performed to estimate the
accuracy and sensitivity. Expected accuracy should be equal
or better than 99%. For clinical applications each test, not
just the technology, should go through clinical trials and be
SNP genotyping
X Chen and PF Sullivan
89
approved by appropriate authorities. The accuracy and
sensitivity must meet the standards before a test is approved.
Success rates are important indexes for high throughput
applications. There are two aspects of success rates. One is
the percentage of markers that can be genotyped by the
technology. The acceptable rate is 80–85% or better. Factors
that contribute to the success rate of markers are the nature
of polymorphisms, allele discrimination and detection
methods. Some mismatches are known to be difficult for
certain allele discrimination, for example, G/T mismatch is
difficult to be distinguished by hybridization and ASE,40
others may be difficult for a detection method, such the A/T
polymorphism for MALDI-TOF MS detection.82,112 Local
sequences where the target polymorphisms located could
limit the choices of primer design this could result in
nonspecific reactions or failures. AS-PCR, the Invader assay
and the TaqMan 50 nuclease assay are more vulnerable than
other methods because they all rely on the immediate local
sequences to design allele-specific primers. Protocols that
use two levels of specificity, namely one for target amplification and the other for allele discrimination, would minimize
the risk of nonspecific reactions. The other success rate is the
percentage of samples that can be genotyped, which is also
referred to as calling rate or scoring rate. Despite the fact
that this measurement is dependent on the quality of DNA
samples, it relates to technologies for their tolerance of
sample quality. It is an important criterion for high
throughput applications because sample quality is likely to
be different and in some cases it is difficult to control for
large-scale, multigroup studies. If a method is too sensitive
to sample quality, it can cause substantial missing data,
which is a major concern for statistical analyses. A preferable
rate would be 95% or better.
From the user’s perspective, flexibility is an important
issue. Users want a technology that can be used for multiple
applications, and that is easy to learn, to design and to scaleup. Flexibility demands a method that is equally efficient
and cost-effective whether it is applied to applications that
require a small number of markers and a large number of
samples or to applications that require a large number of
markers and a small number of samples. Flexibility also
means that the method can be scaled up with similar
efficiency and cost and without further major investment.
The importance of cost and throughput is obvious. For
large-scale applications the cost affects the overall budget
and throughput determines the time table. Only if both the
cost and throughput are reasonable can these applications
become viable.
POPULAR METHODS
In this section, we compare and contrast popular methods
that have been used in both industry and academic
laboratories. We do not intend to cover all techniques,
more comprehensive coverage is available elsewhere.4–6 The
methods listed here were selected based on their simplicity
of principle and protocol, availability of commercial support
and flexibility of setting up to small- and median-sized
laboratories. At the end of this section, we summarize and
compare the features of these methods in Table 1. We hope,
through the discussion, that readers could have a balanced
view of the techniques and make appropriate decisions for
their applications.
SBE with MS Detection
SBE is a robust reaction and there are many platforms for
products detection. The use of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDITOF MS) is one of the platforms promoted by several
companies (Sequenome Inc., http://www.sequenom.com/
genotyping/overview/oview.html, MWG Biotech http://
www.mwgbiotech.com/, and Bruker Doltonics, http://
www.bdal.de/genolink_components.html). In regard to protocol, all begin with PCR amplification, follow with the
clean-up step and the SBE reaction. After SBE reactions, the
samples have to be purified again and to be mixed with
special organic reagents (matrix) and spotted onto a
supporter for MALDI-TOF MS analysis. The differences
between these companies are the ways they prepare DNA
templates and perform SBE reactions. The variations include
modifications made to either or both PCR and extension
primers, the use of tagged ddNTPs or combinations of
ddNTPs and dNTPs, and proprietary matrices that help
crystallization of products for mass detection. These modifications can facilitate product purifications and increase
multiplex capacity. For example, the use of a thiol backbone
in one PCR primer can help preparing single-stranded DNA
templates for primer extension reaction,12 and the use of
mass-code modification for extension primers21 or ddNTPs76
could shift the mass of products to a different range so
multiple extension products can be resolved.
SBE is flexible and can be adapted to many different
applications. SBE has very high success rate (495%). MS
detection is highly sensitive, quantitative and capable of
multiplexing. SBE with MS detection could deliver 10 000–
50 000 genotypes/day, with a consumable cost of $0.15–
0.75/genotype. To achieve this level of throughput, the
method relies on multiplex PCR which requires substantial
optimization and remains largely empirical. Currently, the
multiplex level is 3–5 . The higher throughput of MS
approach makes it more attractive to industry. Another
benefit of MS detection is its quantification. Among the
technologies tested for allele frequency estimation MS
detection appears to be one of the bests.78–82 This feature
becomes more and more important when people realize that
the expected throughput of individual genotyping may not
be easily materialized in near future. In addition to
genotyping, the same equipment can be used for other high
throughput applications such as proteomics and metabolite
screen.113 This feature is very attractive for those users who
work in both fields of functional genomics and proteomics.
MS detection is capable of high throughput. To reach the
potential multiplex PCR is a necessity, but many users may
find it a significant challenge. The relatively higher cost of
initial equipment setup, additional sample handling after
SBE reactions and requirement of special handling of PCR
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
90
Table 1
A summary of popular methods
Method
Strength
SBE with MS detection
High success rate; highly
accurate; quantitative;
capable of multiplexing
SBE with fluorescence
intensity detection
High success rate;
capable of multiplexing
The FP-TDI
Relatively low instrument
cost; simple detection;
easy to automate and
scale-up
Simple protocol; real
time or end point
scoring; easy to automate
and scale-up
The TaqMan 5’ nuclease
assay
Pyrosequencing
Providing sequence
context surrounding the
polymorphism; real-time
scoring; quantitative
The invader assay
No PCR for genotyping;
simple protocol;
isothermal reaction
Weakness
Relatively higher
instrument cost and
longer protocol;
problematic multiplex
PCR
Relatively higher
instrument cost and
longer protocol;
problematic multiplex
PCR
Limited throughput
Very high fixed cost for
each SNP, limited
throughput; requires
optimizations for each
SNP
Lengthy protocol;
relatively expensive
instrument and
consumable cost; limited
throughput
Relatively large amount
of genomic DNA;
requires optimizations for
each SNP; the key
enzyme not available for
general research use
and/or extension primers are potential weakness of MS
methods.
SBE with Fluorescence Detection—the SNP Stream
Genotyping System
The SNP stream system is a method based on SBE with solidphase-mediated detection marketed originally by Orchid
BioSciences.114 At the time this article was written, the
business was sold to Beckman Coulter (http://www.beckman.com/). While those systems sold are still supported by
Orchid BioSciences, the system will be marketed by Beckman Coulter. It relies on multiplex PCR to amplify targets
from several SNPs and after the clean-up of PCR products,
uniquely tagged extension primers for each of the SNPs are
used to perform SBE with fluorescence dye labeled terminators. The unique tag for each SNP serves as a sorting
mechanism to associate the extension products from the
SNP to a specific supporter. The supporter can be a well in a
microplate, an address in a microarray, or a specific
microbead. SBE reactions label the products with allelespecific fluorescence and they are now ready to hybridize to
The Pharmacogenomics Journal
Cost, throughput and
application
Reference
$0.15–0.75/genotype;
10–50k genotypes/day;
most applications
(12, 16–20, 76, 78–82,
95, 119, 120, 129, 130)
$0.2–0.8/genotype;
20–100k genotypes/day;
most applications
(99, 114)
$0.4–0.6/genotype; 1–5k
genotypes/day;
applications with o1
million genotypes
Cost highly dependent
on sample size; 1–5k
genotypes/day; suitable
for applications with fixed
SNPs and large sample
sizes
$1–2/genotype; 1–2k
genotypes/day; suitable
for applications with
pooled samples
(32, 35, 100, 125)
No cost information for
general research; 1–5k
genotypes/day; clinical
applications
(46, 61, 107–109, 133)
(14, 64, 74, 80)
(43, 80, 103, 105, 116,
117, 122, 131, 132)
the supporters. A washing step is necessary to remove
nonspecific hybridization and unreacted terminators. Fluorescence signals can be read out by appropriate systems,
which, depending on the supporters used, can be a
fluorescence plate reader, a flow cytometry, a microarray
scanner or other imaging systems. Genotypes are scored by
the fluorescence intensities and their ratios.
The key feature of the system is the tagged extension
primers. The tags can be made as a universal set. Based on
the use of the tags, Orchid BioSciences recently developed
‘arrays in array’ microplates where each well of the plate is a
small array with 16 addresses. This feature enables it to score
up to 12 SNPs in a well. The system comes with software to
design primers and to score genotypes, and some models
also provide automation tools. The system, which costs
between $250 000 and 500 000, can deliver 20 000–100 000
genotypes/day, and like mass spectrometry detection, its
throughput is largely dependent on the successful implementation of multiplex PCR. Running costs are slightly
higher than that of the MS detection because of the
use of fluorescence terminators and tag arrays. The
SNP genotyping
X Chen and PF Sullivan
91
implementation of multiplex PCR, therefore, is critical
to throughput and cost, without which it is arguably
inferior to FP-TDI method which offers simpler protocol
and detection.
Fluorescence Polarization detection for Template-directed
Dye-terminator Incorporation Assay
This method uses SBE for allelic discrimination and
fluorescence polarization for products detection.32,35 The
method is marketed by Perkin-Elmer corporation (http://
www.nen.com/licensing/genotype.htm).The protocol consists of three sequential reactions—PCR, enzymatic clean-up
or SAP/exonuclease I digestion and SBE. The reaction is
normally performed on a black, skirted PCR plate, in either
96- or 384-well format. When one step is finished, only a
single addition of a reaction mixture to the previous
reaction is needed. There is no sample transferring and no
purification between reactions. When all reactions are
completed, the reaction plate can be loaded directly onto a
FP reader and genotypes are scored for all samples in the
plate in a few minutes. The protocol can be easily automated
with simple liquid handling equipment.
Like SBE with MS detection and the SNP stream system,
FP-TDI is flexible and has high success rate. It can be adapted
for many applications that require moderate amount of
genotyping for 102–103 markers and 102–103 samples. The
main difference between FP-TDI and the other two methods
mentioned above is that FP-TDI does not multiplex PCR and
it scores genotypes directly after SBE reaction without any
manipulation, the other two methods require multiplexing
PCR and need further processing after SBE reaction. If
multiplexing PCR is not successful, the two methods would
have similar throughput as FP-TDI, which is about 1000–
5000 genotypes/day. For FP-TDI the entire protocol can be
viewed as a module, the best way to increase throughput is
to increase the number of modules, which is essentially the
capacity of singular PCR. The consumable cost is about
$0.40–0.60/genotype. Initial instrument cost is relatively
low compared with other methods, starting at $25 000.
The TaqMan 50 Nuclease Assay
The 50 nuclease assay is an elegant assay. It is a closed tube,
single-step assay, and can score genotypes in real time or at
the end of reaction. It combines target DNA amplification
with allele discrimination in a single reaction. The allele
discrimination relies on the hybridization of two doubly
labeled allele-specific fluorescence probes to the target
polymorphisms.14 When the allele-specific probe matches
the target perfectly, they form a stable duplex structure.
During DNA synthesis (PCR), the moving DNA polymerase
will cleave (digest) the doubly labeled allele-specific probe.
The cleavage of the probe changes the FRET status between
the two fluorophores. This change of FRET can be monitored
in real time or as end point assay. If the probe has one base
mismatch (at the polymorphic site) with the target, the
duplex structure formed would not be as stable as duplex
formed by perfectly matched probe. When the DNA
polymerase encounters such easily disrupted duplex, it does
not cleave the probe, instead, it pushes the probe off the
template. This action does not change the FRET status
between the two fluorophores so no signals would be
observed during the reaction.
Since the difference between the two alleles is determined
by the single-base mismatch, the design of the allele-specific
probes is critical for the assay. Despite that the vendor PE
Applied Biosystems worked out some rules to guide the
design, it is still largely empirical and each new SNP has to
go through substantial testing and optimization. The use of
a minor groove binding agent47 has greatly improved its
performance, and makes its optimization much easier. The
assay is currently marketed by PE Applied BioSystems using
FRET detection format. An FP detection version was also
reported.74
From protocol point of view, the TaqMan assay is the best
among the popular methods: it has only one setup step and
does not require any postreaction handling for genotype
scoring. Like other homogeneous assays, the TaqMan assay
has limited capacity for multiplexing and, therefore,
throughput is a bottleneck. The instruments that can be
used for the TaqMan assay are available from several
manufacturers in either 96- or 384-well plate format. The
cost of the instruments is from $50 000 to over $150 000.
The throughput capacity is about 1000–3000 genotypes/
day/machine when the assay is run in 384-well format and
in real-time mode. When the assay is run in end point
mode, which is a little more problematic, the throughput is
substantially higher, about 5 000–10000 genotypes/day.
The major weakness of the method is its fixed cost. The
use of two doubly-labeled fluorescence probes (FRET format)
can be very expensive (somewhere between $200 and 400/
probe) when the sample volume is not large enough to
absorb the fixed cost. The running cost is cheap, basically
the cost of PCR, about $0.15–0.35/genotype. For that reason
applications that use large sample volume and fixed SNP
markers, like those in clinical and diagnostic settings, could
take advantage of the simplicity of its protocol.
Pyrosequencing
Pyrosequencing detects DNA sequences by synchronized
primer extension (Figure 2c). The method is marketed by
Pyrosequencing AB (http://www.pyrosequencing.com). The
method designs extension primer a few bases upstream from
the polymorphic site. During primer extension dNTPs are
added one by one as dictated by the target sequences. If the
incoming base matches the template the base would be
added to the extending primer and a pyrophosphate would
be produced. The pyrophosphate then triggers the synthesis
of ATP, which in turn is used by a luciferase to produce a
chemiluminescence signal.41–43,115 If the base added does
not match the template, the primer would not be extended
and the dNTP is then degraded into dNMP by an apyrase.
Since DNA polymerases do not use dNMP, it would not
interfere with subsequent reactions. The clever design of
pyrosequencing is its careful balance of the enzyme
activities between the DNA polymerases and the apyrases.
Under the assay conditions, DNA polymerases are faster
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
92
than the apyrases, so the polymerases always process the
incoming nucleotide first, but this process is conditioned on
the sequences of the template. So if the nucleotide matches
the template, it will be incorporated onto the primer by the
polymerases, otherwise, it will be left to the apyrases, which
are phosphodiesterases that do not discriminate among the
four bases and do not produce pyrophosphate. From the
description of the mechanism, it would be understood that
the amount of dNTPs added and the time window between
two consecutive additions are important, they have to be
taken into account the balance of the enzymes that compete
to use them.
Pyrosequencing is a unique technique that provides local
sequences surrounding the polymorphic sites. It is a method
other than DNA sequencing that can provide sequence
contexts of target sites. The design of probes is straightforward, and the measurements are quantitative.116,117 However, the assay is an enzymatic cascade involving multiple
enzymes which increase its cost and complexity of the
protocol. Because not all enzymes involved are thermostable, heat cannot be used to denature the templates so the
assay is run at room temperature. This requires preparation
of single-stranded templates, a process that is time consuming and tedious. The low reaction temperature also increases
the chances of nonspecific priming of the primers and the
templates. To facilitate single-stranded template preparation
the current protocol uses a biotinylated PCR primer, which
is significantly more expensive than regular primers. The
throughput is limited by PCR capacity and template
preparation. The pyrosequencer supports 96-well plate
format, the real-time detection takes about 20–30 min for
a 96-well plate. The running cost of the methods is about
$1.0–2.0/genotype and the instrument costs about
$100 000.
Invader Technology
This technology is marketed by the Third Wave Technologies, Inc. (http://www.twt.com) The Invader technology
relies on a structure-specific endonuclease activity found in
DNA polymerases of some archael bacteria,45,118 and it
claims to be the only technology that does not require PCR
amplification. This feature is attractive for clinical applications because PCR license fees for clinical application are
quite expensive. The company markets the core technology,
that is, the biochemistry. The detection comes with multiple
formats, products can be detected by gel electrophoresis,
FRET,91 FP61 and MALDI-TOF MS.119,120 The company does
not market a detection platform, but any equipment that is
capable of gel electrophoresis, FRET, FP or MS detection can
be used to score the results.
The protocol for the Invader assay is simple. After an
optional PCR amplification, a primary reaction mixture
containing the two allele-specific probes and the invader
probe is added to PCR products or genomic DNA and the
samples are denatured and reannealed. Then a secondary
reaction mixture containing the DNA polymerase and
the secondary, signal amplifying universal oligos is added
and the samples are incubated at a constant temperature
The Pharmacogenomics Journal
(60–701C) for 2–4 h. After that the reactions would be ready
for gel, FRET or FP detection. For MS detection, the samples
have to be purified and mixed with a matrix, then spotted
on a support for measurement.
The capability to genotype directly from genomic DNA is
attractive for some applications that do not want to use PCR.
When the assay is combined with limited PCR, it can be very
robust and powerful. A recent report,121 that combines
multiplex PCR and the Invader assay accomplished impressive throughput, seems support this view. In the report,
100 SNPs were multiplexed in PCR and followed with the
Invader assay. The authors reported that they could score
96–98 SNPs out of the 100 SNPs that were multiplexed. As
calculated, this could reach a throughput of 300–400k/day.
The authors did not elaborate the 100 multiplex PCR.
Since the samples were used for the Invader assay, we should
not read too much into the capability of PCR multiplex,
rather we would consider it a supporting evidence of
superior sensitivity of the Invader technology. This is
because the Invader technique is capable of genotyping
directly from the genomic level, a slightly increase of the
targeted fragments over other genomic region, say 10–100s
folds, would significantly improve the performance of the
assay. A multiplex PCR that produces products with a
million-fold difference in concentration would be considered a failure, but that is sufficient to improve the
performance of the Invader assay.
The company currently supports only clinical applications
and limited customer assay development. This excludes the
technology from general research use because the company
does not sell the key enzyme—cleavase separately. The
requirements of many oligonucleotides for each SNP
and special quality of the invader probe increase its fixed
cost and make it problematic for high throughput applications.
DISCUSSION
The technologies for SNP genotyping have been improved
rapidly in the last few years. Merging new technologies pose
challenge for researchers who are not directly involved in
technology development but need to use the technologies
for their researches. In this article, we reviewed current
technologies, focusing on biochemistries for allelic discrimination, mechanisms for product identification and
detection, and structures of cost and throughput. By
considering the principles and the strength and weakness
of the technologies, we hope to provide essential information for selection of technologies appropriate for given
applications.
Technologies are tools to serve applications. Since
researchers have different objectives for their applications
appropriate genotyping methods may differ. In essence, all
technologies can genotype accurately and effectively, the
more salient differences among the technologies are instrument investment, cost and throughput structure and time
and effort needed to train new users. In considering new
technologies and platforms, we should first focus on the
SNP genotyping
X Chen and PF Sullivan
93
requirements of applications. For applications that require a
few millions of genotypes the methods currently available
are very limited, these are SBE with MS and fluorescence
detection. Both methods have flexible design and can
accommodate applications with large number of samples
or large number of markers. Both methods are accurate and
have good success rate. As discussed above, although the
methods are capable of delivering such throughput, the
actual throughput accomplished will be determined by the
capability to perform multiplex PCR, which varies greatly
among different research groups. The use of microarrays
could potentially deliver such throughput, but this is
restricted to the applications that use relatively smaller
sample size. For applications that require less than a million
genotypes, in addition to the two methods mentioned
above, the FP-TDI, the TaqMan assay and pyrosequencing
are competent for the job. While all of the second tier
methods have good tracking records in terms of accuracy
and success rate each of them has its unique features that
attract certain researchers. FP-TDI has a straightforward
design, low instrument cost and fixed cost, but its protocol
is not as elegant as the TaqMan assay and its quantification
feathers are not as good as that of pyrosequencing. For the
Taqman assay, its protocol could be considered the best, but
at the expense of higher instrument and fixed costs. While
pyrosequencing has a lengthy protocol and expensive
running cost, it is one of the bests (the other one is MS
detection) for allele frequency quantification,80,116,117 so it
can be justified when pooled samples are used for genotyping.
The quantification feature of the technologies merits
some discussions. As more and more researchers evaluate
pooled samples genotyping for association and linkage
disequilibrium studies, the quantification feature of the
technologies is likely to become an important factor for
consideration of new technologies. Of the technologies
available today, there are many studies that aim at evaluating the accuracy of allele frequency determination have
been published. Examples, in addition to MS detection and
pyrosequencing mentioned above, include AS-PCR,123 the
TaqMan 50 nuclease assay,80 SBE with DHPLC analysis,122
and SBE with electrophoresis,80,124 and the FP-TDI assay.125
For most technologies, the allele frequencies estimated from
pooled samples deviate 1–5% from that of individually
typed samples. The implication of this error rate to a study
depends on study design, the risk factor of the disease and
pooling strategies. Some researchers are optimistic about
genotyping pooled samples.126–128
In addition to instrument, cost and throughput, time and
efforts needed to train new users should also be considered.
From this point of view the experience and expertise of
trainees and the technical support from the company are
important. It is not uncommon for the same technique to
have different outcomes when it is implemented by
different researchers. Under some circumstances, hiring a
new person with appropriate expertise is more efficient than
training existing staff. Adequate and timely supports from
the vendor (eg, hands-on demonstration and assistance with
trouble-shooting protocols) can make the learning and
implementation easier and more effective.
In conclusion, understanding of the biochemistries and
protocols of genotyping helps users to identify techniques
that are appropriate for them. While there may be several
techniques that can suit the applications, the selection of a
technique has to weigh factors of instrument, cost,
throughput, technical support and staff expertise. A successful implementation of a new technology needs combined
efforts from the user and the vendor.
DUALITY OF INTEREST
XC is one of the inventors of the FP-TDI technique and has a small
interest from sales of the technique by Perkin-Elmer Corporation.
ABBREVIATIONS
SNP
SBE
ASE
AS-PCR
FRET
FP
MALDI
MS
single nucleotide polymorphism
single base extension
allele-specific extension
allele-specific PCR
fluorescence resonance energy transfer
fluorescence polarization
matrix-assisted laser desorption/ionization
mass spectrometry
REFERENCES
1 Kruglyak L. Prospects for whole-genome linkage disequilibrium
mapping of common disease genes. Nat Genet 1999; 22: 139–144.
2 Sachidanandam R, Weissman D, Schmidt SC, Kakol JM, Stein LD, Marth
G et al. A map of human genome sequence variation containing 1.42
million single nucleotide polymorphisms. Nature 2001; 409: 928–933.
3 Venter JC, Adams MD, Myers EW, Li PW, Mural RJ, Sutton GG et al. The
sequence of the human genome. Science 2001; 291: 1304–1351.
4 Syvanen AC. Accessing genetic variation: genotyping single nucleotide
polymorphisms. Nat Rev Genet 2001; 2: 930–942.
5 Kwok PY. Methods for genotyping single nucleotide polymorphisms.
Annu Rev Genom Hum Genet 2001; 2: 235–258.
6 Shi MM. Enabling large-scale pharmacogenetic studies by highthroughput mutation detection and genotyping technologies. Clin
Chem 2001; 47: 164–172.
7 Kirk BW, Feinsod M, Favis R, Kliman RM, Barany F. Single nucleotide
polymorphism seeking long term association with complex disease.
Nucleic Acids Res 2002; 30: 3295–3311.
8 Qi X, Bakht S, Devos KM, Gale MD, Osbourn A. L-RCA (ligation-rolling
circle amplification): a general method for genotyping of single
nucleotide polymorphisms (SNPs). Nucleic Acids Res 2001; 29: E116.
9 Wu DY, Wallace RB. The ligation amplification reaction (LAR)—
amplification of specific DNA sequences using sequential rounds of
template-dependent ligation. Genomics 1989; 4: 560–569.
10 Fors L, Lieder KW, Vavra SH, Kwiatkowski RW. Large-scale SNP scoring
from unamplified genomic DNA. Pharmacogenomics 2000; 1:
219–229.
11 Nikiforov TT, Rendle RB, Goelet P, Rogers YH, Kotewicz ML, Anderson
S et al. Genetic Bit Analysis: a solid phase method for typing single
nucleotide polymorphisms. Nucleic Acids Res 1994; 22: 4167–4175.
12 Sauer S, Lechner D, Berlin K, Plancon C, Heuermann A, Lehrach H et al.
Full flexibility genotyping of single nucleotide polymorphisms by the
GOOD assay. Nucleic Acids Res 2000; 28: E100.
13 Tyagi S, Bratu DP, Kramer FR. Multicolor molecular beacons for allele
discrimination. Nat Biotechnol 1998; 16: 49–53.
14 Livak KJ. Allelic discrimination using fluorogenic probes and the 50
nuclease assay. Genet Anal 1999; 14: 143–149.
15 Chen X, Livak KJ, Kwok PY. A homogeneous, ligase-mediated DNA
diagnostic test. Genome Res 1998; 8: 549–556.
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
94
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Bray MS, Boerwinkle E, Doris PA. High-throughput multiplex SNP
genotyping with MALDI-TOF mass spectrometry: practice, problems
and promise. Hum Mutat 2001; 17: 296–304.
Sun X, Ding H, Hung K, Guo B. A new MALDI-TOF based minisequencing assay for genotyping of SNPS. Nucleic Acids Res 2000;
28: E68.
Abdi F, Bradbury EM, Doggett N, Chen X. Rapid characterization of
DNA oligomers and genotyping of single nucleotide polymorphism
using nucleotide-specific mass tags. Nucleic Acids Res 2001; 29:
E61–E61.
Chen X, Fei Z, Smith LM, Bradbury EM, Majidi V. Stable-isotopeassisted MALDI-TOF mass spectrometry for accurate determination of
nucleotide compositions of PCR products. Anal Chem 1999; 71:
3118–3125.
Griffin TJ, Smith LM. Single-nucleotide polymorphism analysis
by MALDI-TOF mass spectrometry. Trends Biotechnol 2000; 18:
77–84.
Ross P, Hall L, Smirnov I, Haff L. High level multiplex genotyping by
MALDI-TOF mass spectrometry. Nat Biotechnol 1998; 16: 1347–1351.
Ye F, Li MS, Taylor JD, Nguyen Q, Colton HM, Casey WM et al.
Fluorescent microsphere-based readout technology for multiplexed
human single nucleotide polymorphism analysis and bacterial identification. Hum Mutat 2001; 17: 305–316.
Fan JB, Chen X, Halushka MK, Berno A, Huang X, Ryder T et al. Parallel
genotyping of human SNPs using generic high-density oligonucleotide tag arrays. Genome Res 2000; 10: 853–860.
Ferguson JA, Steemers FJ, Walt DR. High-density fiber-optic DNA
random microsphere array. Anal Chem 2000; 72: 5618–5624.
Walt DR. Techview: molecular biology. Bead-based fiber-optic arrays.
Science 2000; 287: 451–452.
Hacia JG, Fan JB, Ryder O, Jin L, Edgemon K, Ghandour G et al.
Determination of ancestral alleles for human single-nucleotide polymorphisms using high-density oligonucleotide arrays. Nat Genet 1999;
22: 164–167.
Wang DG, Fan JB, Siao CJ, Berno A, Young P, Sapolsky R et al. Largescale identification, mapping, and genotyping of single-nucleotide
polymorphisms in the human genome. Science 1998; 280:
1077 –1082.
Gilles PN, Wu DJ, Foster CB, Dillon PJ, Chanock SJ. Single nucleotide
polymorphic discrimination by an electronic dot blot assay on
semiconductor microchips. Nat Biotechnol 1999; 17: 365–370.
Liu Q, Sommer SS. Pyrophosphorolysis-activatable oligonucleotides
may facilitate detection of rare alleles, mutation scanning and analysis
of chromatin structures. Nucleic Acids Res 2002; 30: 598–604.
Liu Q, Sommer SS. Pyrophosphorolysis-activated polymerization
(PAP): application to allele-specific amplification. Biotechniques 2000;
29: 1072–1080.
Sanger F, Nicklen S, Coulson AR. DNA sequencing with
chain-terminating inhibitors. Proc Natl Acad Sci USA 1977; 74:
5463–5467.
Hsu TM, Chen X, Duan S, Miller RD, Kwok PY. Universal SNP
genotyping assay with fluorescence polarization detection. Biotechniques 2001; 31: 560–568.
Syvanen AC. Solid-phase minisequencing as a tool to detect DNA
polymorphism. Methods Mol Biol 1998; 98: 291–298.
Pastinen T, Partanen J, Syvanen AC. Multiplex, fluorescent, solid-phase
minisequencing for efficient screening of DNA sequence variation. Clin
Chem 1996; 42: 1391–1397.
Chen X, Levine L, Kwok PY. Fluorescence polarization in homogeneous
nucleic acid analysis. Genome Res 1999; 9: 492–498.
Lindroos K, Liljedahl U, Raitio M, Syvanen AC. Minisequencing on
oligonucleotide microarrays: comparison of immobilisation chemistries. Nucleic Acids Res 2001; 29: E69.
Pastinen T, Raitio M, Lindroos K, Tainola P, Peltonen L, Syvanen AC.
A system for specific, high-throughput genotyping by allelespecific primer extension on microarrays. Genome Res 2000; 10:
1031–1042.
Myakishev MV, Khripin Y, Hu S, Hamer DH. High-throughput SNP
genotyping by allele-specific PCR with universal energy-transferlabeled primers. Genome Res 2001; 11: 163–169.
Germer S, Higuchi R. Single-tube genotyping without oligonucleotide
probes. Genome Res 1999; 9: 72–78.
The Pharmacogenomics Journal
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Ayyadevara S, Thaden JJ, Shmookler Reis RJ. Discrimination of primer
30 -nucleotide mismatch by taq DNA polymerase during polymerase
chain reaction. Anal Biochem 2000; 284: 11–18.
Ronaghi M, Karamohamed S, Pettersson B, Uhlen M, Nyren P. Realtime DNA sequencing using detection of pyrophosphate release. Anal
Biochem 1996; 242: 84–89.
Ronaghi M, Uhlen M, Nyren P. A sequencing method based on realtime pyrophosphate. Science 1998; 281: 363–365.
Ronaghi M. Pyrosequencing sheds light on DNA sequencing. Genome
Res 2001; 11: 3–11.
Dong F, Allawi HT, Anderson T, Neri BP, Lyamichev VI. Secondary
structure prediction and structure-specific sequence analysis of singlestranded DNA. Nucleic Acids Res 2001; 29: 3248–3257.
Kaiser MW, Lyamicheva N, Ma W, Miller C, Neri B, Fors L et al. A
comparison of eubacterial and archaeal structure-specific 50 -exonucleases. J Biol Chem 1999; 274: 21387–21394.
Mein CA, Barratt BJ, Dunn MG, Siegmund T , Smith AN, Esposito L et
al. Evaluation of single nucleotide polymorphism typing with invader
on PCR amplicons and its automation. Genome Res 2000; 10:
330–343.
Walburger DK, Afonina IA, Wydro R. An improved real time PCR
method for simultaneous detection of C282Y and H63D mutations in
the HFE gene associated with hereditary hemochromatosis. Mutat Res
2001; 432: 69–78.
de Kok JB, Wiegerinck ET, Giesendorf BA, Swinkels DW. Rapid
genotyping of single nucleotide polymorphisms using novel minor
groove binding DNA oligonucleotides (MGB probes). Hum Mutat
2002; 19: 554–559.
Tong J, Barany F, Cao W. Ligation reaction specificities of an NAD(+)dependent DNA ligase from the hyperthermophile Aquifex aeolicus.
Nucleic Acids Res 2000; 28: 1447–1454.
Tong J, Cao W, Barany F. Biochemical properties of a high fidelity DNA
ligase from Thermus species AK16D. Nucleic Acids Res 1999; 27:
788–794.
Sriskanda V, Shuman S. Specificity and fidelity of strand joining by
Chlorella virus DNA ligase. Nucleic Acids Res 1998; 26: 3536–3541.
Nilsson M, Antson DO, Barbany G, Landegren U. RNA-templated DNA
ligation for transcript analysis. Nucleic Acids Res 2001; 29: 578–581.
Pickering J, Bamford A, Godbole V, Briggs J, Scozzafava G, Roe P et al.
Integration of DNA ligation and rolling circle amplification for the
homogeneous, end-point detection of single nucleotide polymorphisms. Nucleic Acids Res 2002; 30: E60.
Thomas DC, Nardone GA, Randall SK. Amplification of padlock probes
for DNA diagnostics by cascade rolling circle amplification or the
polymerase chain reaction. Arch Pathol Lab Med 1999; 123:
1170–1176.
Nilsson M, Barbany G, Antson DO, Gertow K, Landegren U. Enhanced
detection and distinction of RNA by enzymatic probe ligation. Nat
Biotechnol 2000; 18: 791–793.
Cutler DJ, Zwick ME, Carrasquillo MM, Yohn CT, Tobin KP, Kashuk C et
al. High-throughput variation detection and genotyping using microarrays. Genome Res 2001; 11: 1913–1925.
Jacobsen N, Bentzen J, Meldgaard M, Jakobsen MH, Fenger M,
Kauppinen S et al. LNA-enhanced detection of single nucleotide
polymorphisms in the apolipoprotein E. Nucleic Acids Res 2002;
30: E100.
Simeonov A, Nikiforov TT. Single nucleotide polymorphism genotyping using short, fluorescently labeled locked nucleic acid (LNA) probes
and fluorescence polarization detection. Nucleic Acids Res 2002;
30: E91.
Prince JA, Feuk L, Howell WM, Jobs M, Emahazion T, Blennow K et al.
Robust and accurate single nucleotide polymorphism genotyping by
dynamic allele-specific hybridization (DASH): design criteria and assay
validation. Genome Res 2001; 11: 152–162.
Chen X, Kwok PY. Homogeneous genotyping assays for single
nucleotide polymorphisms with fluorescence resonance energy transfer detection. Genet Anal 1999; 14: 157–163.
Hsu TM, Law SM, Duan S, Neri BP, Kwok PY. Genotyping singlenucleotide polymorphisms by the invader assay with dual-color
fluorescence polarization detection. Clin Chem 2001; 47: 1373–1377.
De Angelis DA. Why FRET over genomics? Physiol Genom 1999; 1:
93–99.
SNP genotyping
X Chen and PF Sullivan
95
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
Chen X, Zehnbauer B, Gnirke A, Kwok PY. Fluorescence energy
transfer detection as a homogeneous DNA diagnostic method. Proc
Natl Acad Sci USA 1997; 94: 10756–10761.
Tapp I, Malmberg L, Rennel E, Wik M, Syvanen AC. Homogeneous
scoring of single-nucleotide polymorphisms: comparison of the 50 nuclease TaqMan assay and Molecular Beacon probes. Biotechniques
2000; 28: 732–738.
Dubertret B, Calame M, Libchaber AJ. Single-mismatch detection
using gold-quenched fluorescent oligonucleotides. Nat Biotechnol
2001; 19: 365–370.
Tong AK, Li Z, Jones GS, Russo JJ, Ju J. Combinatorial fluorescence
energy transfer tags for multiplex biological assays. Nat Biotechnol
2001; 19: 756–759.
Ju J, Glazer AN, Mathies RA. Energy transfer primers: a new
fluorescence labeling paradigm for DNA sequencing and analysis.
Nat Med 1996; 2: 246–249.
Ju J, Ruan C, Fuller CW, Glazer AN, Mathies RA. Fluorescence energy
transfer dye-labeled primers for DNA sequencing and analysis. Proc
Natl Acad Sci USA 1995; 92: 4347–4351.
Wilkins SP, Hall JG, Lyamichev V, Neri BP, Lu M, Wang L et al. Analysis
of single nucleotide polymorphisms with solid phase invasive cleavage
reactions. Nucleic Acids Res 2001; 29: E77.
Chen X, Kwok PY. Template-directed dye-terminator incorporation
(TDI) assay: a homogeneous DNA diagnostic method based on
fluorescence resonance energy transfer. Nucleic Acids Res 1997; 25:
347–353.
Whitcombe D, Theaker J, Guy SP, Brown T, Little S. Detection of PCR
products using self-probing amplicons and fluorescence. Nat Biotechnol 1999; 17: 804–807.
Solinas A, Brown LJ, McKeen C, Mellor JM,Nicol J, Thelwell N et al.
Duplex Scorpion primers in SNP analysis and FRET applications. Nucleic
Acids Res 2001; 29: E96.
Gibson NJ, Gillard HL, Whitcombe D, Ferrie RM, Newton CR, Little S. A
homogeneous method for genotyping with fluorescence polarization.
Clin Chem 1997; 43: 1336–1341.
Latif S, Bauer-Sardina I, Ranade K, Livak KJ, Kwok PY. Fluorescence
polarization in homogeneous nucleic acid analysis II: 50 -nuclease assay.
Genome Res 2001; 11: 436–440.
Pusch W, Wurmbach JH, Thiele H, Kostrzewa M. MALDI-TOF mass
spectrometry-based SNP genotyping. Pharmacogenomics 2002; 3:
537–548.
Fei Z, Ono T, Smith LM. MALDI-TOF mass spectrometric typing of
single nucleotide polymorphisms with mass-tagged ddNTPs. Nucleic
Acids Res 1998; 26: 2827–2828.
Pusch W, Kraeuter KO, Froehlich T, Stalgies Y, Kostrzewa M. Genotools
SNP manager: a new software for automated high-throughput MALDITOF mass spectrometry SNP genotyping. Biotechniques 2001; 30:
210–215.
Werner M, Sych M, Herbon N, Illig T, Konig IR, Wjst M. Large-scale
determination of SNP allele frequencies in DNA pools using MALDITOF mass spectrometry. Hum Mutat 2002; 20: 57–64.
Ross P, Hall L, Haff LA. Quantitative approach to single-nucleotide
polymorphism analysis using MALDI-TOF mass spectrometry. Biotechniques 2000; 29: 620–629.
Shifman S, Pisante-Shalom A, Yakir B, Darvasi A. Quantitative
technologies for allele frequency estimation of SNPs in DNA pools.
Mol Cell Probes 2002; 16: 429–434.
Buetow KH, Edmonson M, MacDonald R, Clifford R, Yip P, Kelley J et al.
High-throughput development and characterization of a genomewide
collection of gene-based single nucleotide polymorphism markers by
chip-based matrix-assisted laser desorption/ionization time-of-flight
mass spectrometry. Proc Natl Acad Sci USA 2001; 98: 581–584.
Mohlke KL, Erdos MR, Scott LJ, Fingerlin TE, Jackson AU, Silander K et
al. High-throughput screening for evidence of association by using
mass spectrometry genotyping on DNA pools. Proc Natl Acad Sci USA
2002; 99: 16928–16933.
Lindblad-Toh K, Winchester E, Daly MJ, Wang DG, Hirschhorn JN,
Laviolette JP et al. Large-scale discovery and genotyping of singlenucleotide polymorphisms in the mouse. Nat Genet 2000; 24:
381–386.
Iannone MA, Taylor JD, Chen J, Li MS, Rivers P, Slentz-Kesler KA
et al. Multiplexed single nucleotide polymorphism genotyping by
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
oligonucleotide ligation and flow cytometry. Cytometry 2000; 39:
131–140.
Medintz I, Wong WW, Berti L, Shiow L, Tom J, Scherer J et al. Highperformance multiplex SNP analysis of three hemochromatosis-related
mutations with capillary array electrophoresis microplates. Genome Res
2001; 11: 413–421.
Tabor S, Richardson CC. A single residue in DNA polymerases of the
Escherichia coli DNA polymerase I family is critical for distinguishing
between deoxy- and dideoxyribonucleotides. Proc Natl Acad Sci USA
1995; 92: 6339–6343.
Ward CP, Fensom AH, Green PM. Biallelic discrimination assays for the
three common Ashkenazi Jewish mutations and a common non-Jewish
mutation, in Tay–Sachs disease, using fluorogenic TaqMan probes.
Genet Test 2000; 4: 351–358.
Restagno G, Gomez AM, Sbaiz L, De Gobbi M, Roetto A, Bertino E et
al. A pilot C282Y hemochromatosis screening in Italian newborns by
TaqMan technology. Genet Test 2000; 4: 177–181.
Matsubara Y, Fujii K, Rinaldo P, Narisawa K. A fluorogenic allelespecific amplification method for DNA-based screening for inherited
metabolic disorders. Acta Paediatr Suppl 1999; 88: 65–68.
Giesendorf BA, Vet JA, Tyagi S, Mensink EJ, Trijbels FJ, Blom HJ.
Molecular beacons: a new approach for semiautomated mutation
analysis. Clin Chem 1998; 44: 482–486.
Ledford M, Friedman KD, Hessner MJ, Moehlenkamp C, Williams TM,
Larson RS. A multi-site study for detection of the factor V (Leiden)
mutation from genomic DNA using a homogeneous invader microtiter
plate fluorescence resonance energy transfer (FRET) assay. J Mol Diagn
2000; 2: 97–104.
Patil N, Berno AJ, Hinds DA, Barrett WA, Doshi JM, Hacker CR
et al. Blocks of limited haplotype diversity revealed by highresolution scanning of human chromosome 21. Science 2001; 294:
1719–1723.
Kwok PY. GENOMICS: genetic association by whole-genome analysis?
Science 2001; 294: 1669–1670.
Kwok PY. High-throughput genotyping assay approaches. Pharmacogenomics 2000; 1: 95–100.
Paracchini S, Arredi B, Chalk R, Tyler-Smith C. Hierarchical highthroughput SNP genotyping of the human Y chromosome using
MALDI-TOF mass spectrometry. Nucleic Acids Res 2002; 30: E27.
Taylor JD, Briley D, Nguyen Q, Long K, Iannone MA, Li MS et al. Flow
cytometric platform for high-throughput single nucleotide polymorphism analysis. Biotechniques 2001; 30: 661–669.
Cai H, White PS, Torney D, Deshpande A , Wang Z, Keller RA et al.
Flow cytometry-based minisequencing: a new platform for highthroughput single-nucleotide polymorphism scoring. Genomics 2000;
66: 135–143.
Chen J, Iannone MA, Li MS, Taylor JD, Rivers P, Nelsen AJ et al. A
microsphere-based assay for multiplexed single nucleotide polymorphism analysis using single base chain extension. Genome Res
2000; 10: 549–557.
Armstrong B, Stewart M, Mazumder A. Suspension arrays for high
throughput, multiplexed single nucleotide polymorphism genotyping.
Cytometry 2000; 40: 102–108.
Akula N, Chen YS, Hennessy K, Schulze TG, Singh G, McMahon FJ.
Utility and accuracy of template-directed dye-terminator incorporation with fluorescence-polarization detection for genotyping
single nucleotide polymorphisms. Biotechniques 2002; 32:
1072–1076, 1078.
Swan KA, Curtis DE, McKusick KB, Voinov AV, Mapa FA, Cancilla MR.
High-throughput gene mapping in Caenorhabditis elegans. Genome
Res 2002; 12: 1100–1105.
Straub RE, Jiang Y, MacLean CJ, Ma Y, Webb BT, Myakishev MV et al.
Genetic variation in the 6p22.3 gene DTNBP1, the human ortholog of
the mouse dysbindin gene, is associated with schizophrenia. Am J Hum
Genet 2002; 71: 337–348.
Nordfors L, Jansson M, Sandberg G, Lavebratt C, Sengul S, Schalling M
et al. Large-scale genotyping of single nucleotide polymorphisms by
Pyrosequencingtrade mark and validation against the 50 nuclease
(Taqman((R))) assay. Hum Mutat 2002; 19: 395–401.
Ringquist S, Alexander AM, Rudert WA, Styche A, Trucco M.
Pyrosequence-based typing of alleles of the HLA-DQB1 gene.
Biotechniques 2002; 33: 166–175.
www.nature.com/tpj
SNP genotyping
X Chen and PF Sullivan
96
105
106
107
108
109
110
111
112
113
114
115
116
117
118
Berg LM, Sanders R, Alderborn A. Pyrosequencing technology and the
need for versatile solutions in molecular clinical research. Expert Rev
Mol Diagn 2002; 2: 361–369.
Gharizadeh B, Kalantari M, Garcia CA, Johansson B, Nyren P. Typing of
human papillomavirus by pyrosequencing. Lab Invest 2001; 81:
673–679.
Hessner MJ, Friedman KD, Voelkerding KV, Huber S, Ryan D, Nuccie B
et al. Multisite study for genotyping of the factor II (prothrombin)
G20210A mutation by the invader assay. Clin Chem 2001; 47:
2048–2050.
Agarwal P, Oldenburg MC, Czarneski JE, Morse RM, Hameed MR,
Cohen S et al. Comparison study for identifying promoter allelic
polymorphism in interleukin 10 and tumor necrosis factor alpha
genes. Diagn Mol Pathol 2000; 9: 158–164.
Ryan D, Nuccie B, Arvan D. Non-PCR-dependent detection of the
factor V Leiden mutation from genomic DNA using a homogeneous
invader microtiter plate assay. Mol Diagn 1999; 4: 135–144.
Baron H, Fung S, Aydin A, Bahring S, Jeschke E, Luft FC et al.
Oligonucleotide ligation assay for detection of apolipoprotein E
polymorphisms. Clin Chem 1997; 43: 1984–1986.
Edelstein RE, Nickerson DA, Tobe VO, Manns-Arcuino LA, Frenkel LM.
Oligonucleotide ligation assay for detecting mutations in the human
immunodeficiency virus type 1 pol gene that are associated with
resistance to zidovudine, didanosine, and lamivudine. J Clin Microbiol
1998; 36: 569–572.
Shahgholi M, Garcia BA, Chiu NH, Heaney PJ, Tang K. Sugar additives
for MALDI matrices improve signal allowing the smallest nucleotide
change (A:T) in a DNA sequence to be resolved. Nucleic Acids Res
2001; 29: E91.
Mann M, Hendrickson RC, Pandey A. Analysis of proteins and
proteomes by mass spectrometry. Annu Rev Biochem 2001; 70:
437–473.
Bell PA, Chaturvedi S, Gelfand CA, Huang CY, Kochersperger M, Kopla
R et al. SNPstream UHT: ultra-high throughput SNP genotyping for
pharmacogenomics and drug discovery. Biotechniques 2002; 3:
70–77.
Nyren P, Karamohamed S, Ronaghi M. Detection of single-base
changes using a bioluminometric primer extension assay. Anal
Biochem 1997; 244: 367–373.
Gruber JD, Colligan PB, Wolford JK. Estimation of single nucleotide
polymorphism allele frequency in DNA pools by using pyrosequencing. Hum Genet 2002; 110: 395–401.
Wasson J, Skolnick G, Love-Gregory L, Permutt MA. Assessing
allele frequencies of single nucleotide polymorphisms in DNA
pools by pyrosequencing technology. Biotechniques 2002; 32:
1144–1150.
Lyamichev V, Brow MA, Varvel VE, Dahlberg JE. Comparison of the 50
nuclease activities of taq DNA polymerase and its isolated nuclease
domain. Proc Natl Acad Sci USA 1999; 96: 6143–6148.
The Pharmacogenomics Journal
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
Griffin TJ, Smith LM. Genetic identification by mass spectrometric
analysis of single- nucleotide polymorphisms: ternary encoding of
genotypes. Anal Chem 2000; 72: 3298–3302.
Griffin TJ, Hall JG, Prudent JR, Smith LM. Direct genetic analysis by
matrix-assisted laser desorption/ionization mass spectrometry. Proc
Natl Acad Sci USA 1999; 96: 6301–6306.
Ohnishi Y, Tanaka T, Ozaki K, Yamada R, Suzuki H, Nakamura Y. A
high-throughput SNP typing system for genome-wide association
studies. J Hum Genet 2001; 46: 471–477.
Giordano M, Mellai M, Hoogendoorn B, Momigliano-Richiardi P.
Determination of SNP allele frequencies in pooled DNAs by
primer extension genotyping and denaturing high-performance
liquid chromatography. J Biochem Biophys Methods 2001; 47:
101–110.
Germer S, Holland MJ, Higuchi R. High-throughput SNP allelefrequency determination in pooled DNA samples by kinetic PCR.
Genome Res 2000; 10: 258–266.
Le Hellard S, Ballereau SJ, Visscher PM, Torrance HS, Pinson J, Morris
SW et al. SNP genotyping on pooled DNAs: comparison of genotyping
technologies and a semi automated method for data storage and
analysis. Nucleic Acids Res 2002; 30: E74.
Xiao M, Latif SM, Kwok PY. Kinetic FP-TDI assay for SNP allele
frequency determination. Biotechniques 2003; 34: 190–197.
Pfeiffer RM, Rutter JL, Gail MH, Struewing J, Gastwirth JL. Efficiency of
DNA pooling to estimate joint allele frequencies and measure linkage
disequilibrium. Genet Epidemiol 2002; 22: 94–102.
Barratt BJ, Payne F, Rance HE, Nutland S, Todd JA, Clayton DG.
Identification of the sources of error in allele frequency estimations
from pooled DNA indicates an optimal experimental design. Ann Hum
Genet 2002; 66: 393–405.
Bansal A, van den BD, Kammerer S, Honisch C, Adam G, Cantor CR et
al. Association testing by DNA pooling: an effective initial screen. Proc
Natl Acad Sci USA 2002; 99: 16871–16874.
Fei Z, Smith LM. Analysis of single nucleotide polymorphisms by
primer extension and matrix-assisted laser desorption/ionization timeof-flight mass spectrometry. Rapid Commun Mass Spectrom 2000; 14:
950–959.
Stoerker J, Mayo JD, Tetzlaff CN, Sarracino DA, Schwope I, Richert C.
Rapid genotyping by MALDI-monitored nuclease selection from probe
libraries. Nat Biotechnol 2000; 18: 1213–1216.
Nordstrom T, Alderborn A, Nyren P. Method for one-step preparation
of double-stranded DNA template applicable for use with pyrosequencing technology. J Biochem Biophys Methods 2002; 52: 71–82.
Nordstrom T, Nourizad K, Ronaghi M, Nyren P. Method enabling
pyrosequencing on double-stranded DNA. Anal Biochem 2000; 282:
186–193.
Kwiatkowski RW, Lyamichev V, de Arruda M, Neri B. Clinical, genetic,
and pharmacogenetic applications of the Invader assay. Mol Diagn
1999; 4: 353–364.