Allele-Specific Droplet Digital PCR Combined with a Next-Generation Sequencing-Based Algorithm for Diagnostic Copy Number Analysis in Genes with High Homology: Proof of Concept Using Stereocilin.
Nonetheless, genes or genomic regions with high homology, among others, pose a major challenge to those laboratories because they are not easily amenable to sequencing and copy number analysis with available NGS platforms. Specifically, our group recently showed that the analysis of 193 medically relevant genes, including 85 with strong disease associations such as STRC  (hearing loss), PMS2 (colon cancer), PKD1 (polycystic kidney disease), and SMN1 (spinal muscular atrophy), was stymied by substantial homology issues (13). The challenge is inherent in the fact that NGS reads generated by most available chemistries are not long enough to distinguish long stretches of highly identical DNA sequences, thus leading to false-positive and false-negative variant calls (14).
While certain approaches, such as long-range PCR, can be used to isolate such genes for specific sequencing and potentially copy number analysis, they are neither very simple nor cost-effective and, most importantly, suffer from the lack of scalability. This hinders the use of specific CNV-detection methods as first-tier tests for highly homologous genes. Alternatively, affordable and scalable nonspecific assays can potentially be used as "screening" tools such that only seemingly positive cases are followed up on. One such screening method would ideally be NGS-based, given its scalability, widespread adoption, and the need to homogenize workflows in molecular genetics laboratories.
Here, we describe a clinical algorithm that uses NGS coverage data to infer copy number states at the relevant high homology target(s), followed by confirmatory testing only in the apparently positive patient samples, thus obviating the need for parallel copy number testing on all patients undergoing NGS testing. Although NGS-based algorithms have been published (9-12), their potential application to CNV analysis for genes with high homology would be a novel application, and one of importance to clinical NGS testing, especially because 193 genes that are medically relevant have homologous pseudogenes or regions. Because of the homology issue, these genes are particularly enriched for structural rearrangements and often have recurrent CNVs. It would be impractical to run gene-specific copy number assays for several or all (in a genome-wide testing setting) of those genes on every patient in the clinical laboratory.
Although other techniques, such as multiplex ligation-dependent probe amplification and competitive PCR, can be used (15), here we chose allele-specific droplet digital PCR (ddPCR) as a confirmatory platform. ddPCR has recently been used for several applications, including the accurate detection of CNVs with an unprecedented dynamic range (16). This is largely due to the extreme partitioning of simple TaqMan, usually duplex, PCR assays into thousands of smaller droplet PCR reactions that can be specifically quantified by use of TaqMan probe fluorophores (16). Therefore, ddPCR combines the simplicity of the PCR design, and potentially its allele-specific modified form (17), with high resolution.
As a proof of concept and to illustrate its feasibility and clinical utility, we applied this approach to copy number detection in STRC, a tandemly duplicated gene wherein biallelic pathogenic variants, mainly deletions, significantly contribute to hearing loss (14, 18-20).
Materials and Methods
PATIENT AND CONTROL SAMPLES
Control DNA samples with known STRC copy number status (R-1369, R-1370, R-1371, R-1372) or those genotyped by the Illumina Human 850K Bead Chip (Illumina Inc.) for single-nucleotide polymorphism-based copy number analysis (HL-20, HL-21, HL-22, HL-23) were used. For the latter samples, the data were analyzed by CNV Workshop (21) and vendor-provided analysis software (GenomeStudio). Only unique probes were selected for the analysis of CNVs involving regions that included the STRC and CATSPER2 genes.
The clinical utility of our approach was evaluated by use of DNA from 517 deidentified patients with hearing referred to the Partners Laboratory for Molecular Medicine for OtoGenome testing between April 2015 and November 2016. Patients were clinically consented and samples obtained through physician-ordered genetic testing at the CLIA-accredited partners LMM. The additional follow-up testing and analyses were performed under a waived research consent protocol approved by the Partners Healthcare Institutional Review Board.
DROPLET DIGITAL PCR AND COPY NUMBER ANALYSIS
ddPCR was performed as previously described (12). To the greatest extent possible, we maximized the number of nucleotides that were divergent between STRC and pSTRC in our assay design, especially toward the 3; end of the forward and/or reverse primer sequences (Fig. 1), where mismatches have been shown to have the greatest impact on primer-template binding and overall PCR specificity (22). Two allele-specific ddPCR assays, one targeting exon 23 (STRC.ex23) and one to intron 25 (STRC.in25), were designed. As controls, we used 2 nonspecific ddPCR assays targeting exons 2 (STRC.ex02) and 10 (STRC.ex10). Because CATSPER2 also has substantial homology issues (see Table 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol64/issue4), and to be able to distinguish the STRC/CATSPER2 contiguous deletion, and to diagnose of deafness-infertility syndrome in males, we also designed a CATSPER2 allelespecific ddPCR assay with a similar approach. Here, we focused on STRC, though the CATSPER2 assay had equivalent performance (data not shown).
The following primers and probes were used in the above assays: STRC.ex02, Forward: CACTGCAATT TCTGCCTCCTTC, Probe: CTGAGAGGTAGCCC CG-MGBNFQ, and Reverse: TCTCAGCCTCCCAA GTAGCTG; STRC.ex10, Forward: TGAGGAGCTG CAGAGCCTAGT, Probe: CCTGAGTGATCCAA CG-MGBNFQ, and Reverse: GTCCCATTGGCTG CACATTC; STRC.ex23, Forward: AGAGATGGA GCTCTCAGACTTTG, Probe: AGGACTGCCTG ACATT-MGBNFQ, and Reverse: TTTGCCCATG GCTGCCCGCAG; STRC.in25, Forward: TGCT GAGGTAAAGGTGGACTTACTG, Probe: AGAAG GATCATGAAGGTC-MGBNFQ, and Reverse: AAG GAAGAGATGGCTTCAAATGA; CATSPER2.ex06., Forward: GGCCCACATCTGACATGAGA, Probe: ATGGCCTCTAGAGCAT-MGBNFQ, and Reverse: TGAGCAGCAACATCAAGAGGAA. In addition, the following assay was used as an internal reference: RPP 30, Forward: GATTTGGACCTGCGAGCG, Probe: CTGACCTGAAGGCTCT-MGBNFQ, and Reverse: GCGGCTGTCTCCACAAGT. All STRC and CASTPER2 target probes were labeled with 6-carboxy fluorescein (FAM) while the reference RPP30 probe was labeled with 4,7,2,-trichloro-7,-phenyl-6-carboxyfluo rescein (VIC). All primers and labeled probes were or dered from Integrated DNA Technologies and Life Technologies, respectively.
Data analysis was performed by the QuantaSoft software (Bio-Rad), and the copy number state at each of the STRC and/or pSTRC targets was computed based on the ratio of the concentration of the target (FAM) over the reference (VIC). Copy number calls were made as follows:
Loss = CNV [less than or equal to] 1.5;
No CNV = 1.5-2.5; and
Gain = CNV > 2.5
The OtoGenomeV3 test, performed at the Partners Healthcare Laboratory for Molecular Medicine, involved oligonucleotide-based target capture (Agilent SureSelect) followed by Illumina HiSeq NGS of the coding regions and splice sites of 87 genes associated with nonsyndromic sensorineural hearing loss or syndromes wherein hearing loss was the major presenting feature. This version of the test was updated to include additional probes (n = 79) in the STRC gene and adjacent regions (see Table 2 and Fig. 1 in the online Data Supplement) to support NGS-based copy number calling in this gene. NGS data analysis, quality control, and variant calling was performed as previously described (12).
NGS-BASED COPY NUMBER ANALYSIS
Copy number analysis of NGS data was performed by use of a previously validated inhouse algorithm, VisCap (9, 10, 12). Raw coverage data across each target interval, including the STRC gene, was obtained by the Depth Of Coverage functionality of GATK. Then, fractional coverage values for each target or exon were calculated by dividing target or exon-level coverage values by the net coverage across all targets [i.e., target fractional coverage = (target coverage)/(sum of all target coverages)]. Next, we calculated the [log.sub.2] ratio of each fractional-coverage value divided by the median across the batch (10 samples). [Log.sub.2] thresholds for gains (0.4) and losses (-0.55) were established based on the analysis of positive control samples. This algorithm was shown to have very high analytical sensitivity, while its analytical specificity was significantly improved through visual scoring (10).
COPY NUMBER ANALYSIS IN MEDICALLY RELEVANT GENES WITH HIGH HOMOLOGY
To evaluate the extent to which medically relevant genes with homology issues (n = 193) might require copy number analysis for optimal clinical sensitivity, we estimated the contribution of CNVs to the detection rate in each of the 193 genes. We searched Gene Reviews and documented the percent contribution of CNVs, if available, as well as calculated the proportion of CNVs to the total disease mutations in the Human Gene Mutation Database. Sixty genes had deletions and/or duplications previously reported in patients, and in 40 genes CNVs contributed more than 10% of the all disease-mutation alleles (see Table 1 in the online Data Supplement). This gene number is likely to be an underestimate given that studies to date may not have sufficiently assessed the contribution of CNVs to disease.
STRC ALLELE-SPECIFIC DROPLET DIGITAL PCR ASSAY
STRC encodes the stereocilin protein, a structural component of the sensory hair cells in the inner ear, required for mechanoreception. Biallelic loss of function of the STRC gene is a substantial contributor to nonsyndromic deafness (14, 18-20). Furthermore, a contiguous gene deletion including STRC and an adjacent gene involved in sperm motility, CATSPER2, causes deafness-infertility syndrome in men (23).
STRC is tandemly duplicated, leading to a second nonfunctional copy or pseudogene,pSTRC, with 98.9% homology across the entire locus, 99% identity in the coding sequence, and 100% identity in the first half (exons 1-15) of the STRC gene (14). Therefore, we used the few STRC-unique nucleotides in the second half of the gene, as confirmed by long-range PCR and Sanger sequencing (14), to design an allele-specific ddPCR assay for this gene. A similar approach has been shown to distinguish copy number changes between the 2 highly homologous genes, SMN1 and SMN2 (Bio-Rad).
We tested the performance of the STRC-allele-specific ddPCR assays with 8 different samples known to have 0 (homozygous deletion), 1 (heterozygous deletion), 2 (normal), or 3 (duplication) STRC copies (Table 1). As expected, only the specific assays accurately detected all tested copy number states for STRC (Fig. 1 and Table 1). For example, "4" and "2" copies were detected by the nonspecific, STRC.ex02 and STRC.ex10, assays in the STRC normal and homozygous deletion samples, respectively, demonstrating that these assays were nonspecifically quantifying the 2 copies from each STRC and pSTRC (i.e., 4 total copies). On the other hand, the allele-specific ddPCR assays specifically yielded "2" and "0" STRC copies for the normal and homozygous deletion samples, respectively. It is worth noting that the STRC.ex02 and STRC.ex10 nonspecific assays did not accurately detect the expected "3" and "5" copy number states in the tested heterozygous deletion (2 pSTRC + 1 STRC) and duplication (2 pSTRC + 3 STRC) samples. Instead, 4 total copies were detected in both by the nonspecific assays, most likely suggesting that exons 2 and 10 might not have been part of the STRC deletion or duplication in those 2 samples. Indeed, copy number analysis from the NGS data confirmed this observation (Fig. 2, middle 2 panels). Furthermore, no CATSPER2 copy number change was observed in either sample by the allele-specific ddPCR in this gene (data not shown), consistent with the observation that the nearby first half of the STRC gene (exons 1-15) is not impacted.
In addition to their analytical specificity, the STRC allele-specific ddPCR assays demonstrated high precision across all tested copy number states (Table 1).
NGS-BASED COPY NUMBER ANALYSIS
Using coverage data from NGS, we and others have recently demonstrated the ability to robustly call CNVs (9, 11, 12, 24). We have also validated a highly sensitive algorithm, VisCap, for clinical diagnostic purposes (10). This algorithm uses normalized fractional coverage data at each NGS-targeted region to infer their copy number states (9, 12).
We therefore redesigned our NGS capture for hearing loss genes (25) to include additional probes (n = 79) targeting the STRC gene and its surrounding region (see Fig. 1 and Table 2 in the online Data Supplement). As shown in Fig. 2, all STRC copy number states (0, 1,2, or 3) were appropriately detected by our NGS-based CNV caller and were consistent with the allele-specific ddPCR calls for the same samples (Fig. 1). As expected, calls in the second half of the gene (exons 16-29) were more specific than those in the first half (exons 1-15). This is evident from the copy number calls for the homozygous deletion (Fig. 2, lowest panel), where normalized coverage in exons 1-15 met the threshold for a single copy deletion while in fact this was due to nonspecific sampling of 2 (pSTRC) out of the 4 total copies (STRC + pSTRC). On the other hand, as best demonstrated in the heterozygous deletion and duplication samples (Fig. 2, middle panels), normalized coverage calls in some, but not all, exons between 16-29 showed a specific STRC pattern with 1 copy number deletion (blue) or duplication (red). Nonetheless, some calls in this region were still nonspecific showing, for example, a 3 out of 4 (or 0.75) copy number call, falling below the threshold for a heterozygous deletion (Fig. 2, third panel from top).
Of note, although NGS-based copy number calling in the first half of the STRC gene (exons 1-15) was not specific, it was still extremely useful in assessing the copy number state in this region where no specific ddPCR assays could be designed due to 100% identity with pSTRC. For example, a heterozygous deletion or duplication in STRC exons 1-15 might lead to 3/4 (0.75) or 5/4 (1.25) copy number states, respectively, while a homozygous deletion in this region will show a 2/4 (0.5) seemingly heterozygous call (Fig. 2, lowest panel). Such calls in those exons would at least trigger additional confirmatory testing, such as gel-based analysis, if clinically warranted.
Finally, additional samples (n = 85) were retrospectively and prospectively tested by VisCap and were concordant with the STRC allele-specific ddPCR assays (Table 2 and data not shown).
STRC COPY NUMBER ANALYSIS TESTING ALGORITHM
With a few exceptions, VisCap calls from the STRC-baited regions are not highly specific, rendering those calls unreliable if from a standalone test. To balance this with the potentially prohibitive cost of performing allele-specific ddPCR on all clinical samples, we devised a new STRC copy number testing algorithm. For all patients with hearing loss tested by our NGS panel, copy number analysis in the STRC gene was performed by VisCap. Follow-up testing with the STRC allele-specific ddPCR was then performed for confirmation of positive findings or for cases in which VisCap fails due to coverage issues (Fig. 3). This approach enabled the highly sensitive analytical detection and confirmation of CNVs in the STRC gene and eliminated the need to run the STRC allele-specific ddPCR on the many negative cases in which neither STRC nor pSTRC had a copy number variant (Fig. 2, top panel).
Contribution of STRC Copy Number Analysis to Hearing Loss Genetic Testing Clinical Sensitivity. Using this new algorithm (Fig. 3), we prospectively tested 517 patients with hearing loss referred to our hearing loss NGS diagnostic panel. In total, 122 of 517 were positive, for an overall diagnostic yield of 23.6%. Of those 122 cases, 31 (25.4%) were positive due to biallelic pathogenic variants, mostly deletions, in the STRC gene (Table 2). In other words, STRC copy number detection increased the overall diagnostic yield by 6%, from 17.6% to 23.6% (Fig. 4).
Most of the STRC-positive cases (30 out of 31, or 97%) had at least 1 deletion, with 75% (23 out of 31) carrying 2 deletions (either homozygous or compound heterozygous). Further, 22% (7/31) of cases had 1 deletion and 1 pathogenic sequencing variant, while only 1 case (3%) was positive due to a homozygous pathogenic sequence variant in the STRC gene (Fig. 4). Finally, of the 31 STRC-positive cases, 27 (87%) had mild or moderate hearing loss, which is characteristic for STRC-related deafness (18). Hearing loss severity was not known in the remaining 4 (13%) cases.
Although VisCap has previously been shown to have high analytical sensitivity, its analytical specificity was improved through visual scoring (10). Interestingly, no false-positive calls in the STRC gene were made in our cohort, and only the 30 STRC true-deletion calls were made (Table 2). To further assess analytical sensitivity at the STRC locus, we randomly chose 50 cases with normal VisCap calls in this gene and tested them with our allele-specific ddPCR assays. All 50 cases were confirmed as true negatives.
In addition to the improved positive rate, our testing approach reduced the need for a laborious and costly parallel copy number analysis method by allele-specific ddPCR or any other method in the 486 STRC-negative cases.
CNVs represent a major component of the pathogenic variant spectrum in several disease genes whose sequences share high homology with other genomic regions (see Table 1 in the online Data Supplement). Although essential for diagnostic yield, copy number analysis in those genes is technically challenging and is often not performed in diagnostic laboratories. On the one hand, despite their potential cost-effectiveness, nonspecific approaches are not reliable, need to have wide dynamic range to detect varied (up to 4 or more) copy number states in the homologous region(s), and require follow-up confirmatory testing. On the other hand, parallel methods are highly desirable but are often complex, costly, and, therefore, not easily scalable. To address this, we developed an affordable, highly accurate, and (most importantly) scalable approach for diagnostic copy number analysis in homologous disease genes.
This approach uses coverage data from NGS to infer copy number states at each targeted region, including the STRC gene used as a proof of concept in this study. This is the first application of a clinical NGS algorithm for copy number detection in a pseudogene. Although not highly specific, this method has very high analytical sensitivity (10, this study) and is cost-effective because it only requires running a bioinformatics script on readily generated sequencing data. A disadvantage of the NGS-based analysis is its potentially limited dynamic range in assessing the copy number state of genes/regions that have multiple copies (more than 2) of highly identical sequences. For STRC, we showed that our NGS algo rithm could detect copy number changes in the 0-4 range for the nonspecific probes (Fig. 2). We have not tested situations in which more than 4 copies might exist; however, with appropriately optimized algorithms, a wide dynamic range can be achieved (26). In addition, using a few allele-specific NGS capture probes, we were able to get specific copy number changes in the second half of the STRC gene (Fig. 2). Most NGS algorithms rely on coverage information, which can have inter- and intra-assay variation, thus leading to failed or equivocal outcomes. In our experience, this can be avoided by automating the NGS library preparation and increasing the reference cohort size. Finally, coverage normalization within a batch (n = 10 in our algorithm) might be affected if multiple cases within that run had similar deletions (for STRC in our case), potentially leading to false-positive or false-negative calls. However, use of the median (and not the mean) makes our method more robust and resistant to multiple occurrences of STRC deletions. In addition, given that STRC contributes approximately 6%-10% of the overall diagnostic yield, it will be rare to have more than 1 patient with STRC findings, and even then, it would be very unlikely that 2 patients with STRC findings would have the exact same exon(s) impacted in 1 run.
The second component of our approach relies on ddPCR for confirmation of NGS CNV calls. We took advantage of the very limited unique STRC nucleotides to develop a novel allele-specific ddPCR assay that we showed to be highly specific and reproducible (Fig. 1 and Table 1). Our choice of the ddPCR method as a targeted confirmatory platform was based on its well-documented high analytical sensitivity and wide dynamic range in detecting CNVs (16) and its relatively easier work flow than other methods. In addition, ddPCR employs the TaqMan chemistry, which uses unique nucleotide mismatches, that is amenable to an allele-specific copy number detection form, as we have shown here. However, other methodologies, such as gel-based assays, multiplex ligation-dependent probe amplification, or competitive PCR (15), can also be used if such techniques are more established in a given laboratory.
A potential disadvantage of the allele-specific ddPCR format is its reliance on unique nucleotides to distinguish the true copy number state in the relevant gene. Some regions, such as STRC exons 1-15, might be 100% identical to genomic sequences elsewhere, obviating the design of specific ddPCR assays therein. Different approaches might be required for specific CNV detection in those regions. Alternatively, allele-specific ddPCR assays in flanking regions might be used. One advantage of our approach is its ability to assess, although nonspecifically, copy number changes in high-identity regions through the NGS-based component (Fig. 2).
In summary, we have developed a reliable, scalable, and affordable approach for accurate copy number detection in medically relevant genes with substantial homology issues. Using the STRC gene, we have shown that this approach balances its high analytical sensitivity with its cost-effectiveness and can markedly improve the diagnostic yield of NGS panels that include highly homologous genes.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contribution to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: S.S. Amr, Partners Personalized Medicine Laboratory of Molecular Medicine; H.L. Rehm, Brigham & Women's Hospital.
Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: None declared.
Expert Testimony: None declared.
Patents: None declared.
Role of Sponsor: No sponsor was declared.
Acknowledgment: The authors would like to thank all members of the Partners Laboratory for Molecular Medicine.
(1.) Abou Tayoun AN, Spinner NB, Rehm HL, Green RC, Bianchi DW. Prenatal DNAsequencing: Clinical, counseling, and diagnostic laboratory considerations. Prenat Diagn 2017;37:1-7.
(2.) de Ligt J, Willemsen MH, van Bon BW, Kleefstra T, Yntema HG, Kroes T, et al. Diagnostic exome sequencing in persons with severe intellectual disability. N Engl J Med 2012;367:1921-9.
(3.) Lee H, Deignan JL, Dorrani N, Strom SP, Kantarci S, Quintero-Rivera F, et al. Clinical exome sequencing for genetic identification of rare Mendelian disorders. JAMA 2014;312:1880-7.
(4.) Retterer K, Juusola J, Cho MT, Vitazka P, Millan F, Gibel lini F, et al. Clinical application of whole-exome sequencing across clinical indications. Genet Med 2016; 18:696 -704.
(5.) Talkowski ME, Ordulu Z, Pillalamarri V, Benson CB, Blumenthal I, Connolly S, et al. Clinical diagnosis by whole genome sequencing of a prenatal sample. N Engl J Med 2012;367:2226-32.
(6.) Yang Y, Muzny DM, Reid JG, Bainbridge MN, Willis A, WardP A, et al. Clinical whole-exome sequencing for the diagnosis of Mendelian disorders. N Engl J Med 2013; 369:1502-11.
(7.) Yang Y, Muzny DM, Xia F, Niu Z, Person R, Ding Y, et al. Molecular findings among patients referred for clinical whole-exome sequencing. JAMA 2014;312:1870-9.
(8.) Abou Tayoun AN, Krock B, Spinner NB. Sequencing-based diagnostics for pediatric genetic diseases: progress and potential. Expert Rev Mol Diagn 2016;16:987-99.
(9.) Abou Tayoun AN, Tunkey CD, Pugh TJ, Ross T, Shah M, Lee CC, et al. A comprehensive assay for CFTR mutational analysisusing next-generation sequencing. Clin Chem 2013;59:1481-8.
(10.) Pugh TJ, Amr SS, Bowser MJ, Gowrisankar S, Hynes E, Mahanta LM, et al. VisCap: inference and visualization of germ-line copy-number variants from targeted clinical sequencing data. Genet Med 2016;18:712-9.
(11.) Retterer K, Scuffins J, Schmidt D, Lewis R, Pineda-Alvarez D, Stafford A, et al. Assessing copy number from exome sequencing and exome array CGH based on CNV spectrum in a large clinical cohort. Genet Med 2015;17:623-9.
(12.) Tayoun AN, Mason-Suares H, Frisella AL, Bowser M, Duffy E, Mahanta L, et al. Targeted droplet-digital PCR as a tool for novel deletion discovery at the dfnbl locus. Hum Mutat 2016;37:119 -26.
(13.) Mandelker D, Schmidt RJ, Ankala A, McDonald Gibson K, Bowser M, Sharma H, et al. Navigating highly homologous genes in a molecular diagnostic setting: a resource for clinical next-generation sequencing. Genet Med 2016;18:1282-9.
(14.) Mandelker D, Amr SS, Pugh T, Gowrisankar S, Shakhbatyan R, Duffy E, et al. Comprehensive diagnostic testing For stereocilin: an approach for analysing medically important genes with high homology. J Mol Diagn 2014;16:639-47.
(15.) Zhou L, Palais RA, Paxton CN, Geiersbach KB, Wittwer CT. Copy number assessment by competitive PCR with limiting deoxynucleotide triphosphates and high resolution melting. Clin Chem 2015;61:724-33.
(16.) Hindson BJ, Ness KD, Masquelier DA, Belgrader P, Heredia NJ, MakarewiczAJ, et al. High-throughput drop let digital PCR system for absolute quantitation of DNA copy number. Anal Chem 2011;83:8604-10.
(17.) Heim M, Meyer UA. Genotyping of poor metabolisers of debrisoquine by allele-specific PCR amplification. Lancet 1990;336:529-32.
(18.) Francey LJ, Conlin LK, Kadesch HE, ClarkD, Berrodin D, Sun Y, et al. Genome-wide SNP genotyping identifies the stereocilin (STRC) gene as a major contributor to pediatric bilateral sensorineural hearing impairment. Am J Med Genet A 2012;158A:298 -308.
(19.) Verpy E, Masmoudi S, Zwaenepoel I, Leibovici M, HutchinTP, Del Castillo I, et al. Mutationsina new gene encoding a protein of the hair bundle cause nonsyndromic deafness at the DFNB16 locus. Nat Genet 2001;29:345-9.
(20.) Vona B, Hofrichter MA, Neuner C, Schroder J, Gehrig A, Hennermann JB, et al. Dfnb16 is a frequent cause of congenital hearing impairment: implementation of STRC mutation analysis in routine diagnostics. Clin Genet 2015;87:49 -55.
(21.) GaiX, Perin JC, Murphy K, O'Hara R, D'Arcy M, Wenocur A, et al. CNV Workshop: an integrated platform for high-throughput copy number variation discovery and clinical diagnostics. BMC Bioinformatics 2010;11:74.
(22.) Kwok S, Kellogg DE, McKinney N, Spasic D, Goda L, Levenson C, Sninsky JJ. Effects of primer-template mismatches on the polymerase chain reaction: human immunodeficiency virustype 1 modelstudies. Nucleic Acids Res 1990;18:999-1005.
(23.) Zhang Y, Malekpour M, Al-Madani N, Kahrizi K, Zanganeh M, Lohr NJ, et al. Sensorineural deafness and male infertility: a contiguous gene deletion syndrome. J Med Genet 2007;44:233-40.
(24.) Plagnol V, Curtis J, Epstein M, Mok KY, Stebbings E, Grigoriadou S, et al. A robust model for read count data in exome sequencing experiments and implications for copy number variant calling. Bioinformatics 2012;28: 2747-54.
(25.) Abou Tayoun AN, Al Turki SH, Oza AM, Bowser MJ, Hernandez AL, Funke BH, et al. Improving hearing loss gene testing: a systematic review of gene evidence toward more efficient next-generation sequencing-based diagnostic testing and interpretation. Genet Med 2016;18:545-53.
(26.) Alkan C, Kidd JM, Marques-Bonet T, Aksay G, Antonacci F, Hormozdiari F, et al. Personalized copy number and segmental duplication maps using next-generation sequencing. Nat Genet 2009;41:1061-7.
Sami S. Amr, [1,2] Elissa Murphy,  Elizabeth Duffy,  Rojeen Niazi,  Jorune Balciuniene,  Minjie Luo, [4,5] Heidi L. Rehm, [1,2,3] and Ahmad N. Abou Tayoun [4,5]*
 Laboratory for Molecular Medicine, Partners Healthcare Personalized Medicine, Cambridge, MA;  Department of Pathology, Brigham & Women's Hospital and Harvard Medical School, Boston, MA;  The Broad Institute of MIT and Harvard, Cambridge, MA;  Division of Genomic Diagnostics, The Children's Hospital of Philadelphia, Philadelphia, PA;  The University of Pennsylvania Perelman School of Medicine, Philadelphia, PA.
* Address correspondence to this author at: 3615 Civic Center Blvd., Abramson Research Building, 716E, Philadelphia, PA 19104. Fax 215-590-2156; e-mail aboutayoua@ chop.edu.
Received August 12,2017; accepted December 8,2017.
Previously published online at DOI: 10.1373/clinchem.2017.280685
 Nonstandard abbreviations: NGS, next-generation sequencing; SNVs, single-nucleotide variants; CNVs, copy number variants; ddPCR, droplet digital PCR.
 Human genes: STRC, stereocilin; PMS2, PMS1 homolog 2, mismatch repair system component; PKD1, polycystin 1, transient receptor potential channel interacting; SMN1, survival of motor neuron 1, telomeric; CATSPER2, cation channel sperm associated 2; SMN1, survival of motor neuron 1, telomeric; SMN2, survival of motor neuron 2, centromeric.
Caption: Fig. 1. STRC allele-specific droplet digital PCR assay design and performance.
Top, an STRC gene schematic showing all exons (1-29) as vertical rectangles. Light pink rectangles represent exons 1-15 with 100% identity to the pseudogene pSTRC, whereas exons 16-29 (light green) contain very few bases that are divergent from pSTRC. Filled pink and green circles represent nonspecific (STRC.ex02 and STRC.ex10) and STRC gene-specific (STRC.ex23 and STRC.in25) probes, respectively. Dashed areas highlight regions with divergent bases (red bars) where the specific probes/primers (middle) were designed. The STRC-specific bases within the probe/primer sequences are shown in red in the middle table. Bottom, unlike the nonspecific assays (light pink), the STRC.ex23 and STRCin25 (green) assays accurately detected all tested copy number states (homozygous deletion or "0" copies, heterozygous deletion or "1" copy, Normalor "2" copies, and duplication or"3" copies) in STRC. Copy number states are calculated based on the ratio of the concentration of the target over the reference in each sample (see Methods). Each bar graph represents copy number average [+ or -] SE (n = 2-8 runs). See also Table 1.
Caption: Fig.2. NGS-based copy number analysis in STRC.
Filled circles represent relative copy number changes, as inferred from [log.sub.2] ratios of the fractional coverage value for each baited region divided by the median for that region across the batch (see Methods). Blue circles represent deletions while red circles represent copy number gains. As in Fig. 1, the 4 samples with the following STRC genotypes were tested (from top to bottom): Normal, duplication, heterozygous deletion, and homozygous deletion. Each row represents data from 1 sample with copy number changes across all chromosomes shown on the left, while a zoomed-in view of the STRC region on chromosome 15 (dashed area) is shown on the right. The x axis displays chromosome numbers (left) or STRC-baited region/exon numbers (right).
Caption: Fig. 3. STRC copy numberanalysis algorithm.
See text for additional information.
Caption: Fig. 4. Contribution of STRC copy number analysis to the clinical sensitivity of hearing loss genetic testing.
Left, breakdown of the overall testing results in a cohort of 517 hearing loss patients. The inconclusive category refers to reports with variants of uncertain significance or one heterozygous pathogenic variant in a recessive gene. Only benign or likely benign variants were detected in the "negative" category. "Positive - Other" refers to reports with positive findings in hearing loss genes other than STRC. "Positive - STRC' refers to positive reports due to biallelic pathogenic variants in STRC. Right, of the total 31 positive cases due to STRC, 15 had homozygous deletions, 8 were compound heterozygous for 2 different deletions, 7 were compound heterozygous for a deletion and an SNV, and only 1 had a homozygous pathogenic SNV. See Table 2 for additional genotype information.
Table 1. Concordance and reproducibility of the STRC.ex23 and STRC.in25 allele-specific ddPCR STRC assays. Copy number--Run 1 Sample (a) Previous genotype STRC.ex23 STRC.in25 R-1371-RP1 Normal (CN = 2) 1.80 2.00 R-1371-RP2 1.90 1.90 Mean ([+ or -] SE) 1.850 1.95 (0.05) (0.05) R-1369-RP1 Homozygous loss (CN =0) 0.0 0.00 R-1369-RP2 0.03 0.00 Mean ([+ or -] SE) 0.002 0 (0.002) (0.00) R-1370-RP1 Heterozygous loss (CN = 1) 1.02 0.99 R-1370-RP2 0.97 0.98 Mean ([+ or -] SE) 0.99 0.99 (0.035) (0.005) R-1372-RP1 Duplication (CN = 3) 3.10 3.00 R-1372-RP2 2.90 3.00 Mean ([+ or -] SE) 3.00 3.00 (0.10) (0.00) HL20-RP1 Normal (CN = 2) 1.86 2.03 HL20-RP2 1.8 1.99 2.08 Mean ([+ or -] SE) 1.83 2.01 (0.03) (0.02) HL21-RP1 Homozygous loss (CN =0) 0.002 0.001 HL21-RP2 0.0007 0.0007 Mean ([+ or -] SE) 0.005 0.00 (0.00) (0.00) HL22-RP1 Heterozygous loss (CN = 1) 0.95 0.9 HL22-RP2 0.98 0.86 Mean ([+ or -] SE) 0.97 0.88 (0.01) (0.02) HL23-RP1 Duplication (CN =3) 3.18 2.92 HL23-RP2 2.91 2.72 Mean ([+ or -] SE) 3.05 2.82 (0.13) (0.10) Copy number--Run 2 Sample (a) Previous genotype STRC.ex23 STRC.in25 R-1371-RP1 Normal (CN = 2) 1.90 2.00 R-1371-RP2 2.00 2.10 Mean ([+ or -] SE) 1.95 2.05 (0.05) (0.05) R-1369-RP1 Homozygous loss (CN =0) 0.00 0.00 R-1369-RP2 0.00 0.00 Mean ([+ or -] SE) 0.00 0.00 (0.00) (0.00) R-1370-RP1 Heterozygous loss (CN = 1) 0.92 0.95 R-1370-RP2 0.94 0.94 Mean ([+ or -] SE) 0.93 0.95 (0.01) (0.005) R-1372-RP1 Duplication (CN = 3) 3.00 3.10 R-1372-RP2 2.90 3.00 Mean ([+ or -] SE) 2.95 3.05 (0.05) (0.05) HL20-RP1 Normal (CN = 2) 2 2 HL20-RP2 1.8 2.03 2.01 Mean ([+ or -] SE) 2.04 2.02 (0.04) (0.01) HL21-RP1 Homozygous loss (CN =0) 0 0 HL21-RP2 0 0 Mean ([+ or -] SE) 0 0 (0.00) (0.00) HL22-RP1 Heterozygous loss (CN = 1) 0.93 1 HL22-RP2 0.96 0.97 Mean ([+ or -] SE) 0.95 0.99 (0.01) (0.01) HL23-RP1 Duplication (CN =3) 3.01 3.09 HL23-RP2 3 2.93 Mean ([+ or -] SE) 3.01 3.01 (0.00) (0.08) Copy number--Run 3 Sample (a) Previous genotype STRC.ex23 STRC.in25 R-1371-RP1 Normal (CN = 2) 2.10 1.90 R-1371-RP2 1.90 2.00 Mean ([+ or -] SE) 2.00 1.95 (0.10) (0.05) R-1369-RP1 Homozygous loss (CN =0) 0.00 0.00 R-1369-RP2 0.00 0.00 Mean ([+ or -] SE) 0.00 0.00 (0.00) (0.00) R-1370-RP1 Heterozygous loss (CN = 1) 0.96 0.94 R-1370-RP2 0.98 0.96 Mean ([+ or -] SE) 0.97 0.95 (0.01) (0.01) R-1372-RP1 Duplication (CN = 3) 3.00 3.00 R-1372-RP2 3.10 3.00 Mean ([+ or -] SE) 3.05 3.00 (0.05) (0.00) HL20-RP1 Normal (CN = 2) 2.07 2.03 HL20-RP2 1.8 2.02 Mean ([+ or -] SE) 2.04 2.03 (0.03) (0.01) HL21-RP1 Homozygous loss (CN =0) 0 0.01 HL21-RP2 0 0 Mean ([+ or -] SE) 0 0.001 (0.00) (0.00) HL22-RP1 Heterozygous loss (CN = 1) 0.94 0.93 HL22-RP2 0.93 0.93 Mean ([+ or -] SE) 0.93 0.93 (0.00) (0.00) HL23-RP1 Duplication (CN =3) 3.37 2.82 HL23-RP2 3.12 3.09 Mean ([+ or -] SE) 3.25 2.95 (0.12) (0.13) Sample (a) Previous genotype Concordant R-1371-RP1 Normal (CN = 2) Yes R-1371-RP2 Mean ([+ or -] SE) R-1369-RP1 Homozygous loss (CN =0) Yes R-1369-RP2 Mean ([+ or -] SE) R-1370-RP1 Heterozygous loss (CN = 1) Yes R-1370-RP2 Mean ([+ or -] SE) R-1372-RP1 Duplication (CN = 3) Yes R-1372-RP2 Mean ([+ or -] SE) HL20-RP1 Normal (CN = 2) Yes HL20-RP2 1.8 Mean ([+ or -] SE) HL21-RP1 Homozygous loss (CN =0) Yes HL21-RP2 Mean ([+ or -] SE) HL22-RP1 Heterozygous loss (CN = 1) Yes HL22-RP2 Mean ([+ or -] SE) HL23-RP1 Duplication (CN =3) Yes HL23-RP2 Mean ([+ or -] SE) (a) In each run, all samples were run as duplicates. RP = repeat Table 2. STRC pathogenic alleles in hearing-loss-positive cases. Allele 1 (a) Allele 2 (a) STRC (ex1-26) and CATSPER2 (ex1-7) del STRC (ex1-26) and CATSPER2 (ex1-7) del STRC (ex1-29) del STRC (ex1-29) del STRC (ex16-29) del STRC (ex16-29) del STRC (ex1-26) and CATSPER2 (ex1-7) del STRC (ex16-29) del STRC (ex16-29) del STRC (ex16-25) del STRC (ex16-29) del STRC (ex1-29) and CATSPER2 (ex1-13) del STRC (ex1-26) and CATSPER2 (ex1-7) del c.5188C>T; p.Arg1730* STRC (ex1-26) and CATSPER2 (ex1-7) del c.4027C>T; p.Gln1343* STRC (ex1-26) and CATSPER2 (ex1-7) del c.3670C>T; p.Arg1224* STRC (ex1-26) and CATSPER2 (ex1-7) del c.1086C>A; p.Tyr362* STRC (ex1-26) and CATSPER2 (ex1-7) del c.3217C>T; p.Arg1073* STRC (ex16-29) del c.379C>T; p.Arg127* STRC (ex16-29) del c.4701 + 1G>A; p.? c.3493C>T; p.Gln1165* c.3493C>T; p.Gln1165* Allele 1 (a) Number of patients STRC (ex1-26) and CATSPER2 (ex1-7) del 11 STRC (ex1-29) del 1 STRC (ex16-29) del 3 STRC (ex1-26) and CATSPER2 (ex1-7) del 6 STRC (ex16-29) del 1 STRC (ex16-29) del 1 STRC (ex1-26) and CATSPER2 (ex1-7) del 1 STRC (ex1-26) and CATSPER2 (ex1-7) del 1 STRC (ex1-26) and CATSPER2 (ex1-7) del 1 STRC (ex1-26) and CATSPER2 (ex1-7) del 1 STRC (ex1-26) and CATSPER2 (ex1-7) del 1 STRC (ex16-29) del 1 STRC (ex16-29) del 1 c.3493C>T; p.Gln1165 * 1 (a) Minimum CNV breakpoints are displayed.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Molecular Diagnostics and Genetics|
|Author:||Amr, Sami S.; Murphy, Elissa; Duffy, Elizabeth; Niazi, Rojeen; Balciuniene, Jorune; Luo, Minjie; Reh|
|Date:||Apr 1, 2018|
|Previous Article:||Measurement of Lipoprotein-Associated Phospholipase A2 by Use of 3 Different Methods: Exploration of Discordance between ELISA and Activity Assays.|
|Next Article:||Advanced Whole-Genome Sequencing and Analysis of Fetal Genomes from Amniotic Fluid.|