High accuracy genotyping directly from genomic DNA using a rolling circle amplification based assay

Background Rolling circle amplification of ligated probes is a simple and sensitive means for genotyping directly from genomic DNA. SNPs and mutations are interrogated with open circle probes (OCP) that can be circularized by DNA ligase when the probe matches the genotype. An amplified detection signal is generated by exponential rolling circle amplification (ERCA) of the circularized probe. The low cost and scalability of ligation/ERCA genotyping makes it ideally suited for automated, high throughput methods. Results A retrospective study using human genomic DNA samples of known genotype was performed for four different clinically relevant mutations: Factor V Leiden, Factor II prothrombin, and two hemochromatosis mutations, C282Y and H63D. Greater than 99% accuracy was obtained genotyping genomic DNA samples from hundreds of different individuals. The combined process of ligation/ERCA was performed in a single tube and produced fluorescent signal directly from genomic DNA in less than an hour. In each assay, the probes for both normal and mutant alleles were combined in a single reaction. Multiple ERCA primers combined with a quenched-peptide nucleic acid (Q-PNA) fluorescent detection system greatly accellerated the appearance of signal. Probes designed with hairpin structures reduced misamplification. Genotyping accuracy was identical from either purified genomic DNA or genomic DNA generated using whole genome amplification (WGA). Fluorescent signal output was measured in real time and as an end point. Conclusions Combining the optimal elements for ligation/ERCA genotyping has resulted in a highly accurate single tube assay for genotyping directly from genomic DNA samples. Accuracy exceeded 99 % for four probe sets targeting clinically relevant mutations. No genotypes were called incorrectly using either genomic DNA or whole genome amplified sample.


Background
Sequencing of the human genome has led to the identification of mutations and single nucleotide polymorphisms (SNPs) that can be linked with specific phenotypes. A number of assays have been developed to genotype known SNPs and mutations. However, the majority of these methods have many steps and are difficult to automate. Currently, most genotyping methods require that the target region of the genomic DNA be amplified using PCR prior to the genotyping assay (reviewed in [1,2]). Following amplification, the PCR product is either sequenced directly, or probed for the mutation of interest. Similarly, OCP ligation/ERCA can be used to accurately genotype PCR products in a biallelic format, where the probes for both alleles are present in a single reaction [3]. The OCP ligation/ERCA method is based on ligation dependent circularization of allele specific DNA probes followed by exponential rolling circle amplification of the circularized probes [4,5]. ERCA probes have also been variously described as 'padlock probes' [6,7] or 'circularizable probes' [4]. Although OCP ligation/ERCA can be used to genotype PCR products, it also provides a means for directly genotyping genomic DNA without requiring an initial amplification of the genomic DNA locus. For example, we recently described a high throughput ligation/ERCA protocol for genotyping genomic DNA samples in microtiter plates [8].
This report describes modifications to OCP ligation/ERCA genotyping that reduce reaction time and improve accuracy to above 99%. Methods for improving accuracy include simultaneous detection of two alleles per well, probes designed to contain hairpin sequences that regulate ligation discrimination and prevent nonspecific amplification, and introduction of a Q-PNA fluorescent detection system [9]. Improvements in probe design, rolling circle reaction design, and assay methods were integrated into an improved high accuracy genotyping assay, and the utility of the method was demonstrated for four clinically important mutations. The Q-PNA fluorescent detection system produced a six-fold reduction in time to result.
In many instances only limited quantities of genomic DNA are available from clinical samples for genetic testing. DNA can also be limiting when large numbers of assays must be performed on a single sample. Recently, a whole genome amplification method (WGA) has been developed that is capable of accurately producing large quantities of genomic DNA from limited sample [10,11]. In cases where genomic DNA was limiting, WGA genomic DNA was substituted for genomic DNA. Complete and faithful amplification of the entire genome resulted in WGA product that could be substituted for genomic DNA without affecting the accuracy of genotyping. WGA DNA amplification and OCP ligation/ERCA can be combined to create an automated high throughput mutation assay that includes DNA amplification from limiting samples and allele detection in a single tube.

Probe Design
The OCP ligation/ERCA assay was improved to meet the requirements for fast, and accurate, genotyping. Probes for both alleles were combined in a single reaction tube ( Figure 1). One OCP annealed perfectly to the normal sequence of interest, and the other targeted the mutant. The 3' terminal nucleotide of the OCP annealed opposite the variant nucleotide. If the OCP and the target matched at the mutation site, the OCP was circularized by ligase. If the OCP was mismatched at the 3' end, it was not ligated. There are a number of advantages to having both probes present in the same reaction, as opposed to having each probe in a separate reaction [3]. For instance, if amplification fails in a biallelic (ie., two allele) reaction, signal is not generated for either allele, and the result can be discarded. However, if only one of the two reactions fail in a two reaction/ single allele per well format, a heterozygote could be miscalled as a homozygote. Furthermore, the two allele per well format reduced non-specific amplification. With both mutant and normal OCP in the reaction, either one or the other or both OCPs were circularized, depending on the genotype. Therefore, there was a specific rolling circle reaction in every assay, regardless of the genotype. The kinetics of the specific rolling circle reaction were sufficiently rapid so that non-specific amplification was greatly reduced or eliminated. Details of the mechanism of the Exponential Rolling Circle Amplification (ERCA) reaction are shown in Figure 2.
The exponential amplification of the ligated probe sequence generates double stranded DNA products that are unit lengths of the probe [4]. Electrophoretic separation on an agarose gel generates a ladder of DNA bands as seen in Figure 3. Typically, the products ranged in size from 80 bases to 1.2 kb, with most products less than 500 bases long. The ERCA products for normal and mutant alleles incorporate a different fluorescent P1 primer. In Figure 3, signal from the normal allele results in FAM (false color green) fluorescence, while signal from the mutant allele results in Cy3 (false color red) fluorescence. A sample with heterozygous genotype produces both colors.
The OCPs were redesigned to incorporate a hairpin structure ( Figure 4) that improved target specificity analogous to beacon [12] probe design. Parameters affecting the stability of the hairpin were tailored to increase the specificity of target recognition. Target directed hybridization using a hairpin is more specific than hybridization of a linear probe because the structure of the hairpin probe is maintained in a folded state unless the hairpin sequence encounters its preferred target sequence. If the probe does not encounter the correct target, the hairpin will remain self-annealed, reducing the possibility of nonspecific or transient annealing events. Only the extensive base pairing between the probe and target sequence is sufficient to disrupt the hairpin.
The new probe design also provides a way to reduce background by rendering unused probe inert. ( Figure 5). During ERCA, the 3'-end of the hairpin of unused probe is extended by DNA polymerase to produce a non-reactive double-stranded product (data not shown). In the absence of a method such as this for either removing or inactivating unused OCP, nonspecific background amplification can occur [13]. Sequencing of misamplification products (data not shown, [14]) has revealed that misamplification products produced in the absence of ligase usually consist of a short OCP-derived sequence immediately flanked by primer sequences. Each 'primer-OCP-primer' unit was repeated dozens or hundreds of times in tandem. The unit sequence varied randomly from reaction to reaction, and sometimes more than one sequence was present in a reaction. The exact mechanism for generating and expanding misamplification products with this overall sequence motif is not completely understood. However, improved probe design that results in enzymatic elimination of unused OCPs, as demonstrated with the 3' hairpin OCP, can reduce or eliminate unwanted nonspecific amplifications.
To determine the effect of the 3' hairpin on genotyping accuracy, probes with and without 3' hairpins were designed for both the Factor V Leiden and Hemochromatosis H63D mutations (Table 1). Control probes lacking hairpins were constructed in two different ways: 1) Hairpins in the FVL probes were removed by simply deleting the non-target directed OCP sequence involved in hairpin formation. This did not change priming sites or the base pairing between the probe and target, but non-hairpin probes were reduced in length by 10 nucleotides compared to the hairpin probes. 2) To eliminate the 3' hairpin without changing the length of the OCP, the hairpins in the H63D probes were removed by changing the sequence of the OCP such that the length of the probes was conserved but the base pairing in the hairpin was disrupted. Real time ERCA genotyping reactions were performed

Figure 1
Ligation of OCP to genomic DNA target Open circle probe (OCP) anneals to genomic DNA target, and the 3' base either base pairs with the target (red) or creates a mismatch (blue). The matched OCP is circularized by ligase, and becomes topologically linked to the target.  Figure 6). Results for non-hairpin H63D probes were similar, with only 65% called cor-

Figure 2
ERCA using Q-PNA, P1 and P2 Exponential amplification of circularized OCP was performed using primers P1 and P2. P1 is a composite of a Q-PNA annealed to a fluorescent DNA primer specific for the circle sequence directed against one allele. Q-PNA anneals to the fluorescent DNA primer such that the c-terminal quencher on the PNA quenches the 5' fluorescent moiety (either FAM or Cy3) on the DNA primer. P2 is a DNA primer with the same strand orientation as the circle, complementary to the product of the primary rolling circle reaction. Amplification of the circle specific for only one of the two alleles is illustrated. ERCA begins when P1 anneals to the circle (1), and polymerase copies the circle sequence, eventually copying the entire circle. The polymerase begins displacing the previously synthesized product, opening up single stranded P2 binding sites (2). Each turn of displacement synthesis around the circle exposes another P2 binding site, and the resulting synthesis from the annealed P2 primers results in the displacement of downstream primers and products, including the PNA (3). Displacement of the PNA separates the fluorescent label on the DNA from the quencher on the PNA, producing fluorescent signal specific for the P1/circle pair (4). Displacement of the downstream primers and products also opens up additional P1 binding sites, and the process continues in an exponential cascade. Figure 8). In both cases, the non-hairpin genotyping reactions showed a dramatic increase in the incidence of late appearing, non-specific signal. When the hairpin containing OCPs were used, all 96 samples were genotyped correctly (Figures 7 and 9).
To increase the accuracy of genotyping, probes for both alleles were combined in a single tube format. For heterozygotes, the two probes are concurrently amplified in the ERCA reaction. Exponential amplification of circularized OCP requires primers P1 and P2 (see Figure 2). The concurrent amplification of mutant and normal probes is kept in balance by using the same P2 for each probe. P1 primers were routinely titrated to determine optimum concentration [4]. This approach greatly reduces the difficulty of balancing the kinetics of biallelic ERCA amplifications. If the P2 primer is common to both OCPs, only P1 priming efficiencies need to be similar, greatly simplifying the process of obtaining balanced reaction kinetics in a biallelic assay.
The use of several P1 and P2 primers, which initiate DNA synthesis from multiple positions on the probe and product, increases ERCA rate, sensitivity and specificity ( [15], data not shown). Therefore, secondary amplification primers were designed for each probe. In genotyping reactions where two probes were present in a single reaction, the secondary primers were designed to anneal to target specific sequences shared by both OCPs, in the 5' and/or 3' arms (Table 1).

Genotyping Accuracy
The improvements to the OCP ligation/ERCA genotyping reaction outlined above were incorporated into four gen-otyping assays, and a study was performed to test the reaction design for genotyping accuracy. Probes for the Factor V Leiden (FVL) mutant and normal alleles (see Table 1) were designed using the parameters outlined in Methods. These probes were used to screen a 216 patient sample set that had been previously genotyped using RFLP. The sample groups were enriched for the mutant genotypes that normally occur at very low frequency in the population.  Figure 10. The genotype for each sample was determined

Figure 4
Specific annealing of 3' hairpin OCP to its genomic target followed by ligation. The hairpin structure can increase OCP discrimination during ligation by providing a two state equilibrium for the OCP, where OCP is either in the hairpin form or annealed to target. A mismatch between target and OCP favors the intra-molecular hairpin formation. A match between the OCP and the target provides an extended region of complementarity that is sufficient to maintain stable binding of the 3' end of the OCP to the target, allowing ligase to circularize the OCP. This balance contributes to accurate and sensitive allele discrimination.  Table 2.
Probes for the Factor II prothrombin (FII) mutant and normal alleles (see Table 1) were designed as discussed in Methods. These probes were used to screen a 298 patient sample set that was previously genotyped using RFLP. There were two mutant genotypes, 12 heterozygotes, and 284 homozygous normal genotypes. Because mutant patient samples were limiting, a separate sample group that consisted of multiple repeats of the same Factor II mutant and heterozygous samples was screened. As with the genotyping assays described above, Factor II prothrombin (FII) mutant genotyping results were also visualized either in real time or as an end-point assay ( Figure  11). Summary of results is shown in Table 2. Samples probed for the Factor II mutation were genotyped with 99.5% accuracy within a single assay, 100% accuracy if performed in triplicate.
Probes for the two Hemochromatosis mutations H63D and C282Y were designed using the parameters outlined in Methods and used to screen patient sample sets of known genotype (probe sequences in Table 1). There were 138 samples screened for H63D and 190 samples screened for C282Y. This sample group was also enriched for the mutant genotype, containing 89 homozygous normal, 48 heterozygous and 1 homozygous mutant (the single mutant was repeated 29 times) for the H63 allele and 84 normal, 93 heterozygous and 13 homozygous mutant patient samples for the C282 allele. Examples of typical genotyping results for H63D are shown in Figure 12 and results for C282Y are shown in Figure 13. Genotyping results were visualized as both real time reaction traces and as an end-point assay. Samples screened for both hemochromatosis mutations were genotyped with 99.5% accuracy within a single assay, 100% accuracy if performed in triplicate (Table 2).

Discussion
OCP ligation/ERCA can be used to directly probe genomic DNA for SNPs or mutations, and the entire process can be 3' Hairpin OCPs design allows conversion of unused OCPs to inert double-stranded form. If the OCP is not circularized, the single stranded region of the unligated linear probe can bind P1 primer and be copied, a process that can lead to misamplification of unligated probe. To reduce the potential for misamplification, a 3' hairpin has been designed into the OCP. During ERCA reaction, 3' hairpin is extended by polymerase, resulting in an extended hairpin with minimal single strandedness, which is less likely to participate in faulty priming reactions that result in misamplification. performed in a single tube. The ERCA amplification reaction is rapid, generating signal in as little as 10 minutes, and is incubated at a single temperature. These characteristics make OCP ligation/ERCA easily adaptable to high throughput automated genotyping platforms. Accurate genotyping was obtained in screens designed to detect four clinically relevant mutations, Factor V Leiden, Factor II prothrombin, Hemochromatosis C282Y and Hemochromatosis H63D.
Several improvements to the ligation/ERCA method [8] have increased accuracy to levels acceptable for diagnostic applications and reduced reaction time. Other reports using ERCA based genotyping require PCR amplification of the locus of interest prior to genotyping [3,16], which is prohibitively expensive, time consuming, and more difficult to automate. We have designed ERCA primers that are optimized for minimal misamplification and artifact formation. With any exponential signal amplification method, nonspecific amplification due to exponential artifacts presents a potential problem. In ERCA as in PCR, amplification of primer-primer artifacts can mimic specific signal if primers are not selected carefully. To avoid this problem, commercially available primer design software was used to design P1 and P2 primers that are optimal for amplification. This step does not necessarily have to be repeated for each target, however, which is an advantage over PCR based methodologies. The primary amplification primer sequences are present in the OCP backbone, not in the portion of the OCP that anneals to target, allowing optimized P1 and P2 pairs, to be used for many different targets. In principle, a small collection of optimized P1/P2 pairs should contain primers that can be used for any given target.
The Q-PNA detection system [9] was used as a fluorescent reporter during ERCA. During ERCA, Q-PNA is rapidly displaced from the ERCA product and is physically separated from the fluor in a bimolecular reaction, resulting in detectable signal in as little as 10 minutes. By comparison, Amplifluors typically required 60-120 minutes to generate signal [8]. The substantial difference in time to generate signal may be due to the nature of Amplifluor design. In an Ampliflour, complementary DNA sequences, fluorescent reporter, and quencher are all covalently linked on the same DNA strand. The comparative

Figure 6
OCPs without hairpins, Factor V Leiden Mutation Ligation/ERCA Assay. Optimal assay conditions were as described in Methods. 1000 ng of MDA genomic DNA was incubated with probes for both alleles in the presence of ligase. ERCA mix was added and the reaction was followed in real time at 60°C in a Bio-Rad I-Cycler. The PCR baseline subtracted relative fluorescence units are shown as a function of time for both FAM (top row) and Cy3 (bottom row). Each trace (colored lines) corresponds to a different sample. Real time traces from reactions containing the same genotypes (either normal, heterozygous, or homozygous mutant) were grouped together according to genotype in each of the three columns. Genotypes were assigned as described in Methods.

Figure 7
Hairpin OCPs, Factor V Leiden Mutation Ligation/ERCA Assay-. Assay was performed as described in Figure 6.

HOMOZYGOUS MUTANT HETEROZYGOUS
stability of the unimolecular Amplifluor hairpin is likely to hinder displacement synthesis when compared to the bimolecular Q-PNA system. In addition, the high local concentration of fluor and quencher may serve to supress signal from the Amplifluor. The Q-PNA based reporter system consistently produces signal far more rapidly than the Amplifluor based system, allowing time to result of less than an hour for the entire assay.
Secondary allele specific, nonfluorescent, primers can also be used to increase the speed and specificity of ERCA. These primers significantly advanced the rate of ERCA amplification, cutting the reaction time in half for the Factor V Leiden probe set (data not shown), resulting in a 20minute assay. Although maximum fluorescent signal decreased slightly as each additional nonfluorescent primer was added, the addition of one or two more primers to the reaction did not adversely influence signal strength to the extent that genotyping was compromised.
Each of the four assays developed for this report contained two probes, one for each allele. Using both probes in a single reaction has been demonstrated to reduce the levels of nonspecific amplification due to primer dimers and misamplification of unligated probe [3]. At least one of the probes will always be amplified, suppressing low levels of nonspecific signal.
The introduction of a hairpin into the design of the open circle probes, similar to approaches taken in molecular beacon design [12], provides a means to regulate the degree of ligation discrimination. A 3' and/or 5' stem-loop structure may be designed to increase specific binding to difficult sequences. The ratio of the hairpin stability to target annealing stability directly affects the overall target annealing specificity. By modulating the stability of the hairpin any OCP can be fine tuned for ligation specificity.
The OCP 3' hairpin also decreases the level of nonspecific DNA synthesis caused by misamplification of OCP sequences. Analysis of misamplified products has shown that OCP sequences are usually found in the misamplified DNA products. This means that uncircularized OCP is able to provide a template for primer-based misamplification. Removing the unligated OCP helps to reduce background amplification [13], but digestion of the unused OCP or purification of the ligated circles introduces time consuming and expensive steps. Designing a 3' hairpin into the OCP creates a self-priming sequence in unligated OCP. During ERCA, the polymerase will extend unligated OCP starting at the 3' end in a suicide pathway that renders the OCP double stranded and inert. This approach eliminates nonspecific amplification and improves genotyping accuracy without the introduction of separate isolation or purification steps. As demonstrated above, removal of the hairpin sequence either by

Figure 9
Hemochromatosis H63D Mutation Ligation/ERCA Assay-Hairpin OCPs. Assay was performed as described in Figure 6. Additional methods to improve ERCA specificity were integrated into the genotyping protocol. Careful design allowed the same P2 sequence to be used in the ERCA reaction for both alleles. For each set of OCPs, the P2 sequence was generic. The generic P2 was not part of the target specific portion of the probe, which allowed it to be used in both the normal and mutant OCP. Because both probes in each assay contained the same P2, the reaction kinetics could be more easily balanced, promoting uniform amplification for both alleles. Uniform amplification is especially important when the genotype is heterozygous so that both alleles are accurately represented.  Genotyping accuracy depends on using sufficient quantities of genomic DNA target. As expected, OCP ligation/ ERCA results were influenced by the amount of genomic DNA used in the assay. Typically, genotyping was possible with as little as 50 ng target DNA. However, specific signal strength increased with more DNA target, up to 1000 ng, after which no further benefit was seen (data not shown). WGA DNA can be produced in milligram quantities from nanograms [10] of original sample allowing use of high DNA concentration for optimal signal. As a result, 1000 ng WGA DNA was used to obtain optimal genotyping reaction conditions. The results obtained for the WGA DNA agreed in every instance with the known genotypes determined by RFLP. WGA product should also be compatible with other genotyping assays where better results can be obtained using more target DNA.
Many aspects of probe design are open to manipulation, and in practice probes can be successfully designed for most target sequences of interest. Pickering et al. demonstrated a 95% success rate in designing OCPs for 99 different targets on the first attempt [3]. Probes can be designed against either strand of genomic DNA, and the backbone portion can also be varied. Optimization of T m parameters, and sequence allows rapid design of OCPs with a high degree of success.
The degree to which a genotyping assay needs to be optimized depends in part on the purpose of the assay. A genotyping system intended for diagnostic purposes needs to meet the highest standards of accuracy, robustness, and throughput, as compared to a system that is intended solely for research purposes. In recognition of the expectation that 100% accuracy is a requirement, even under suboptimal conditions, the assays developed for the targets in this report have undergone extensive optimization to maximize accuracy even under less than optimal conditions. As a result, the concentrations of primers, probes and polymerase vary between assays. After determining the best conditions for each individual genotyping assay, the expectation is that these conditions would be the best possible for genotyping large numbers of samples with high accuracy. The critical nature of the information derived from the output of the assay justifies the extra optimization effort. To minimize the potential for error, the accuracy for a single assay was compared to the accuray for the same assay performed in triplicate. Depending on the assay requirements, the level of accuracy can be selected by increasing or decreasing the number of repeats. The demonstration that high levels of input genomic DNA improve robustness of genotyping results is paralleled by PCR results. WGA can be used to generate sufficient DNA for genotyping using large quantities of template. The cost of performing WGA is offset by the fact that sample prep is unnecessary; WGA product can be performed directly on crude blood sample and yields enough DNA for dozens or hundreds of genotyping reactions [11]. Under different circumstances, it may be necessary to rapidly generate assays for a large number of SNPs. For this, the emphasis could be placed on optimization of assay design. Amounts of input DNA, primers and polymerase could be standardized, rapidly yielding accurate genotyping assays but without the highest accuracy under suboptimal conditions. The process of OCP ligation followed by ERCA has been designed such that genomic DNA can be genotyped directly in less than an hour. Hundreds of individual samples were genotyped with over 99% accuracy. Accurate genotyping was demonstrated for both genomic DNA and whole genome amplified DNA. Isothermal exponential amplification produces a fluorescent signal from ligated OCP using nanomolar amounts of probe. Because OCP ligation/ ERCA does not rely on thermal cycling for amplification, amplification kinetics are not limited by cycling rate. The exponential nature of the reaction produces millions of fold amplification in as little as 20 minutes. Output can be read on a fluorescent plate reader, real time PCR instrument, fluorescent imager, or other device equipped with the ability to measure fluorescent signals. OCP ligation/ ERCA probes can be multiplexed. OCP ligation and ERCA reactions for each SNP target are performed in a single

Hairpin OCP Design
All oligomers were ordered from Integrated DNA Technologies, Inc.(Coralville, IA). Sequences of oligomers are shown in Table 1. All OCPs were 5' phosphorylated and gel purified. OCPs were designed such that the 3' terminal nucleotide annealed opposite the mutation of interest ( Figure 1). The T m for the 3' arm of the probe annealed to target was designed to be between 60-65°C. The 5' arm of all OCPs annealed to target upstream of the mutation and was designed to have a T m of approximately 70°C. Oligo To reduce or eliminate the artifactual amplification of unligated OCPs, a hairpin was designed into the 3' end of the OCP, with approximately 10 bases in the stem and 10 in the loop. The hairpin was calculated to be stable at 60°C in the context of the entire OCP using the DNA folding server at Michael Zucker's website http://bioinfo.math.rpi.edu/~zukerm/ OCP sequence was chosen such that the hairpin was the only stable structure in the OCP, with a ∆G of -0.5 to -1.5 kcal/mol.

Generic P1 and P2 design
ERCA primers P1 and P2 were chosen using Oligo 6. These were generic primer sequences, and could be used to amplify any circle sequence. First, 10-20 kb of randomly generated DNA sequence was scanned for primers using settings limiting 3' dimer ∆G to -1.0 kcal/mol, 3' dimer length to 3, 3' terminal stability range -5.5 to -8.0 kcal/mol, GC clamp stability to -10 kcal/mol, no acceptable loop, T m range 52.1 to 56.2°C. Compatible primers were selected using the multiplex function of Oligo 6. A subset of these compatible primers was utilized as generic P1 and P2 primers (see Table 1). ERCA P1 primers were

Figure 11
Factor II Prothrombin Mutation: 32 repeats of 3 patient samples, 3 different genotypes. Assay was performed as described in Figure 10.

Allele Specific ERCA primers
The sequence of these primers was chosen from the target specific 5' and 3'-arms of the OCP, either complementary to the OCP (5'-arm) or the exact sequence of the OCP (3'arm). These primers annealed to both the normal and mutant OCPs. The sequences of these allele specific primers are shown in Table 1. Allele specific primers were desalted after synthesis.

Q-PNA
Exponential amplification of circularized OCP (see Figure  2) was accomplished using generic primers P1 and P2, along with allele specific primer(s). Each P1 primer contained a universal 13 base 5'-tail (see Table 1) with a fluorescent dye at the 5'-end. Fluorescence was quenched in unincorporated primers by the annealing of a 13-residue PNA molecule, Q-PNA-13 (Applied Biosystems/Boston Probes, Bedford, MA). The sequence of Q-PNA-13 is Ac-X-OO-TGA-TTG-CGA-ATG-A-Lys (Dabcyl). The C-terminal dabcyl was positioned so that it was proximal to the 5'-fluorescent moiety on the DNA reporter, quenching fluores-

Figure 12
Hemochromatosis H63D Mutation: 71 different patient samples, 3 different genotypes. Assay was performed as described in Figure 10. End Point Assay cence when the reporter and Q-PNA-13 were annealed, as shown in Figure 2. After P1 primer is incorporated into ERCA product, Q-PNA is displaced, producing a fluorescent signal.

Genomic DNA
Genomic DNA samples were purchased from Coriell Cell Repositories, Camden, NJ. Human Variation Panel, Caucasian panel of 100, Reference # HD100CAU, was used as non-mutant controls and for general screening purposes. Human samples containing specific mutations were also purchased from Coriell: Factor V Leiden mutation (heter-ozygous NA14642 and homozygous NA14899); Hemochromatosis C282Y (heterozygous NA14642, homozygous NA14620); and Hemochromatosis H63D (heterozygous NA14641 and homozygous NA13591); Factor II Prothrombin mutation (heterozygous NA16028 and homozygous NA16000). Other patient DNA samples were genotyped using RFLP, and an aliquot of the purified DNA was subsequently used either directly in genotyping reactions, or used after Whole Genome Amplification (WGA).

Figure 13
Hemochromatosis C282Y Mutation: 96 different patient samples, 3 different genotypes. Assay was performed as described in Figure 10.

Wild Type Heterozygous
Mutant minutes, depending on the assay). After the reaction was complete, the results could be read at any time. Some reactions were allowed to remain at room temperature overnight, some were stored at 4°C, some were frozen at -20°C. In each case, the reaction was protected from light to prevent photobleaching of the fluorescent reporters.
The genotype of each sample was determined automatically using a modified fuzzy c-means clustering algorithm [3], which groups the data into three genotypes plus a negative control, and assigns a confidence level to each genotyping call from 0 (not in cluster) to 1 (100% certainty that point belongs to cluster).