An efficient screening method for simultaneous ...

Clinica Chimica Acta 445 (2015) 34–40

Contents lists available at ScienceDirect

Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim

An efficient screening method for simultaneous detection of recurrent copy number variants associated with psychiatric disorders Julio Rodriguez-Lopez a,b, Noa Carrera a,b,1, Manuel Arrojo a,c, Jorge Amigo a,d, Beatriz Sobrino b, Mario Páramo a,c, Eduardo Paz a,c, Santiago Agra a,c, Ramón Ramos-Ríos a,c, Julio Brenlla a,c, Ángel Carracedo a,b,d, Javier Costas a,⁎ a

Instituto de Investigación Sanitaria (IDIS) de Santiago de Compostela, Servizo Galego de Saúde (SERGAS), Santiago de Compostela, Spain Fundación Pública Galega de Medicina Xenómica, Santiago de Compostela, Spain Servizo de Psiquiatría, Complexo Hospitalario Universitario de Santiago de Compostela, Servizo Galego de Saúde (SERGAS), Santiago de Compostela, Spain d Grupo de Medicina Xenómica, Universidade de Santiago de Compostela, Santiago de Compostela, Spain b c

a r t i c l e

i n f o

Article history: Received 15 August 2014 Accepted 5 March 2015 Available online 20 March 2015 Keywords: Copy number variants Quantitative interspecies competitive PCR Psychiatric genetics Schizophrenia Autism

a b s t r a c t Several recurrent copy number variants (CNVs) increasing risk to neuropsychiatric diseases have been identified in recent years. They show variable clinical expressivity, being associated with different disorders, and incomplete penetrance. However, due to its very low frequency, the full variety of clinical outcomes associated with each one of these CNVs is unknown. Current methods for detection of CNVs are labor intensive, expensive or not suitable for high throughput analysis. Quantitative interspecies competitive PCR linked to variant minisequencing and detection by mass-spectrometry may overcome these limitations. Here, we present two multiplex assays based on this method to screen for eleven psychiatric risk CNVs, such as 1q21, 16p11.2, 3q29, or 16p13.11 regions, among others. The assays were tested in our collection of 514 schizophrenia patients. Results were compared with MLPA at two CNVs. Additional positive results were confirmed by exome sequencing. A total of fourteen patients were CNV carriers. The method presents high sensitivity and specificity, showing its utility as a cheap, accurate, high throughput screening tool for recurrent CNVs. The method may be very useful for management of psychiatric patients as well as screening of different collections of samples to better identify the full spectrum of clinical variability. © 2015 Elsevier B.V. All rights reserved.

1. Introduction One of the most important recent discoveries in psychiatric genetics was the existence of several recurrent submicroscopic microduplications and microdeletions, known as copy number variants (CNVs), that confer risk to several different neurodevelopmental disorders such as autism spectrum disorders, schizophrenia, intellectual disability, or generalized epilepsy [1–3]. Kirov et al. [4] have estimated the overall penetrance of these CNVs for schizophrenia, autism spectrum disorders, developmental delay and congenital malformations ranging from 10.6% to 100%. Most of these CNVs are highly deleterious, being removed by purifying selection in less than 5 generations. Its frequency in populations is very low, generally less than 0.05%, established by an equilibrium between mutation rate and purifying selection [5]. ⁎ Corresponding author at: Instituto de Investigación Sanitaria (IDIS) de Santiago de Compostela, Hospital Clínico Universitario, edificio Consultas, andar -2, Grupo de xenética psiquiátrica, despacho 15, E-15706 Santiago de Compostela, Spain. Tel.: +34 981955452; fax: +34 981951473. E-mail address: [email protected] (J. Costas). 1 Present address: Medical Research Council Centre for Neuropsychiatric Genetics and Genomics, Institute of Psychological Medicine and Clinical Neurosciences, Cardiff University, Cardiff CF24 4HQ, UK.

http://dx.doi.org/10.1016/j.cca.2015.03.013 0009-8981/© 2015 Elsevier B.V. All rights reserved.

Currently, there are at least 14 recurrent CNVs associated with schizophrenia risk, according to a recent meta-analysis [6]. Only one of them, the 22q11.2 deletion responsible for the velocardiofacial and diGeorge syndromes, was known for several years. The identification of these rare CNVs has been possible mainly due to technical improvements in hybridization-based technologies along the genome, and because of the high mutation/recurrence rate of these CNVs [7,8]. Analysis of different collections indicated that the combined frequency of these confirmed schizophrenia risk CNVs is around 2.5% of schizophrenic samples [6]. Several PCR-based approaches are available for detection of specific CNVs, such as multiplex-ligation dependent amplification (MLPA), quantitative real-time PCR (qPCR), multiplex amplicon quantification (MAQ), or invader assay, among others [9]. Unfortunately, each of these methods is labor intensive, requiring careful optimization of primers, probes and/or reaction conditions. In addition, some of them are relatively expensive and sensible to DNA quality. Therefore, there is an urgent need for the development of robust methods allowing the efficient, accurate, and cheap measurement of these CNVs in clinical samples with neurodevelopmental disorders. One of the alternative methods is quantitative competitive PCR, a technique based on amplification of a test sequence in the presence of a known quantity of a competitor sequence that differs from the

J. Rodriguez-Lopez et al. / Clinica Chimica Acta 445 (2015) 34–40

test sequence by a single nucleotide. This method has the advantage of easy multiplexing. The use of chimpanzee DNA as a competitor precludes the need of synthetic DNA for each tested loci [10]. In this work, we present a quantitative interspecies competitive PCR (qicPCR) design, linked to minisequencing and variant detection by mass spectrometry, as a useful method to test for several recurrent CNVs involved in neurodevelopmental disorders at a reduced cost and high accuracy in large number of samples. We applied this method to our collection of schizophrenic patients, identifying an excess of these CNVs in comparison to expected population frequencies.

2. Material and methods 2.1. Samples A total of 514 schizophrenic samples were included in the study. The samples are from the Santiago de Compostela healthcare area (Galicia, NW Spain) and meet the DSM-IV criteria for schizophrenia. All samples gave their written informed consent for this study. The study was performed in accordance with the latest version of the Declaration of Helsinki and has been approved by the Galician Ethical Committee for Clinical Research. Further details are presented in Carrera et al. [11]. Pan troglodytes DNA from the cell line EB176 (JC) was provided by the Health Protection Agency Culture Collection (UK). This sample corresponds to the chimpanzee used for the generation of the reference genome.

35

2.2. Selection of CNVs CNVs for screening were selected from bibliography, mainly, the review of Malhotra and Sebat [3] and the meta-analysis of Levinsson et al. [2]. VIPR2 region was selected based on Vacic et al. [12]. CNVs conferring higher risk to develop schizophrenia were prioritized in the group selected to carry out our study.

2.3. Quantitative interspecies competitive PCR Identification of nucleotide positions with non-conserved nucleotides at human and chimpanzee reference sequences was done by inspection of the alignments of orthologous regions corresponding to the CNVs at the ENSEMBL web server (http://www.ensembl.org). Nucleotide positions were selected if there was no other difference in human–chimpanzee comparison in 100 bp at each edge. The positions were preferentially located in exons of different genes at each region. A minimum of five nucleotide positions per CNV region were chosen. These positions were used as input to design primers for PCR as well as for minisequencing using the Sequenom (Sequenom, Inc.; San Diego, California) MassArray Assay Design Suite v1.0 (https:// seqpws1.sequenom.com/AssayDesignerSuite.html). Five ng of chimpanzee DNA and 5 ng of DNA from each human sample were subject to competitive PCR, following by detection of the amplification products by mass spectrometry using the Sequenom MassArray technology, according to manufacturer's instructions. Analysis of results was based on the height of peaks corresponding to human or chimpanzee non-conserved nucleotides. First, quality

qicPCR-Sequenom Two subsets of samples, n=51 and n=463 (514)

11 CNVs studied

relCN2 SD

qicPCR-Sequenom Replication

124 samples 11 CNVs studied relCN2 SD

WES confirmation, Exome Depth 9 samples with SZ risk CNVs 4 CNVs studied

Fig. 1. Study design. A total of 514 samples were analyzed for the presence of 11 copy number variants (CNVs) by qicPCR using Sequenom MassArray. Those 14 samples with relCN N 2 SD from the mean (n = 124) were subject to a second round of analysis by qicPCR-Sequenom. Those samples whose relCN were again N 2 SD from mean were considered as positive results. Whole exome sequencing (WES) of the 9 positive samples with CNVs with established risk in schizophrenia (SZ), were used to confirm presence of CNVs. MLPA results for the whole sample were used as additional control for both positive as well as negative results at CNV regions 15q13.3 and 16p11.2.

36

J. Rodriguez-Lopez et al. / Clinica Chimica Acta 445 (2015) 34–40

control procedures were performed to remove low-quality assays, those that failed in more than 50% of the samples, as well as low-quality samples, those that failed in more than 50% of the assays. Then, two different normalization procedures were done, similarly to MLPA analysis. First, for each sample, the height of each amplification product was divided by the sum of heights of all peaks in that sample to calculate the relative height. By this way, the remaining CNV regions are considered control regions for each CNVs. Second, for each amplification product, the relative height of the amplification product at each sample was divided by

the sum of the relative heights of this product in all samples. The normalized values were used to estimate the relative amplification of each human sequence in every assay (hsDNArel) as a ratio between the normalized human sequence height and the sum of the normalized human and chimpanzee sequence heights for that assay [10]. Then, the hsDNArel at each CNV region was calculated as the average of all the assays of that region. The relative value of each CNV region per sample (relCN) was estimated as the ratio between the hsDNArel of a specific CNV region against the hsDNArel of the rest of CNV regions from the

Table 1 CNV regions and assays analyzed in the present study. Deletion a

Region

Assays

15q13.3

CHRNA7_extr MTMR15_1 OTUD7A_1 OTUD7A_2 TRPM1 C16orf54_ext MAPK3_1 MAZ PPP4C TMEM219 BCL9_8 CHD1L_9 CHD1L-20-21 FMO5-7_ext PRKAB2_ext BDH_ext1 BDH_ext2 DLG1_2 TFCR_ext1 TFCR_ext2 VIPR2_1 VIPR2_1_2 VIPR2_3 VIPR2_6 VIPR2_7 CYFIP1-001_3-2 CYFIP1-001_6 NIPA1_3utr_2 NIPA1_3utr_4 TUBGCP5_5 NDE1_2-3 NDE1_8 ABCC6_21 ABCC6_29-30 C16orf45_3-4 CDRT4_1utr1 CDRT4_4 COX10_7utr HS3ST3B1_2_2 PMP22_5_2 AATF_9-10 DHRS11_1 HNF1B_8-7 MYO19_14 SYNRG_11 CLDN5 SCARF2_ THAP7_1 ZDHHC8_6 CLTCL1_2-1 C16orf72_1 C16orf72_1utr C16orf72_2 C16orf72_3-4.3 C16orf72_4

16p11.2

1q21.1

3q29

VIPR2 (7q36.3)

15q11.2

16p13.11

17p12

17q12

22q11.2

C16orf72

a b c

Duplication

Neurodevelopmental disorder penetranceb % (95% CI)

Frequency reference samplesc

Carriers in our collection

Neurodevelopmental disorder penetranceb % (95% CI)

Frequency reference samplesc

Carriers in our collection

40 (21–72)

0.00019

–

9.8(3.2–30)

0.000187

2/514

31(19–52)

0.000412

–

34 (22–57)

0.000301

4/514

40 (20–78)

0.000208

–

21 (11–39)

0.000375

–

71 (20–100)

0.000014

–

100 (5.9–100)

0

–

N.A.

0

–

N.A.

0.000685

1/514

13 (9.6–17)

0.002775

2/514

N.A.

N.A.

1/514

15(5.8–40)

0.000392

1/514

10.6 (7–17)

0.001342

2/514

N.A.

0.000259

–

N.A.

0.000240

1/514

43 (14–100)

0.000054

–

19 (5.6–61)

0.000224

–

100 (60–100)

0

–

14(8–24)

0.0007474

–

N.A.

N.A.

–

N.A.

0.000197

–

Assays are named by genic location. Those assays in italics were discarded due to bad performance. Based on Kirov et al. [4]. Based on the largest sample size between the studies of Rees et al. [6], Kirov et al. [4], and Malhotra and Sebat [3].

J. Rodriguez-Lopez et al. / Clinica Chimica Acta 445 (2015) 34–40

37

same individual. Those samples departing more than 2 SD from the mean of each average relCN distribution, corresponding to the different CNV regions, were selected for replication. Samples that deviate again by more than two SD from mean in the replication step were considered CNV carriers (Fig. 1).

the Netherlands) that includes 16 probes for two of the CNV regions tested in our design, 15q13.3 (5 probes) and 16p11.2 (11 probes). The analysis was performed following manufacturer's instructions.

2.4. MLPA

Agilent SureSelect v4 or v5 (Agilent Technologies, California, USA) was used to capture exonic regions of the ten carriers of pathogenic CNVs by in solution hybridization. Captured sequences were sequenced using Solid 5500xl (Life Technologies, Foster City, CA), following

As a control method for CNV detection, we used the MLPA design SALSA MLPA KIT P343-C1 AUTISM-1 from MRC-Holland (Amsterdam,

2.5. Exome sequencing

Fig. 2. Histogram representation of relCN distribution. (a) Ratio of relCN distribution of 16p13.11 region of the SZ224 sample. (b) Ratio of relCN distribution at each one of the assays used to interrogate the 16p13.11 region of the SZ224 sample. The relCN values are shown on the x axis. Dashed lines indicate 2SD from mean. Solid lines indicate the relCN value at the region (a) or each one of the assays (b).

38

J. Rodriguez-Lopez et al. / Clinica Chimica Acta 445 (2015) 34–40

manufacturer instructions. Mapping to the human reference sequencing was performed with the LifeScope pipeline(Life Technologies, Foster City, CA) using default parameters. In addition, the mapped reads generated by LifeScope were used as input for GATK software [13]. The recommended GATK best practices were followed, including detection of duplicated reads by Picard tools (http://picard.sourceforge.net), local realignment around indels and base quality recalibration to obtain the final sequence alignment data. 2.6. CNV detection by exome depth coverage ExomeDepth [14] was used to confirm CNVs at each sample. ExomeDepth is based on the comparison of depth of coverage between the examined sample and an optimized reference set, to identify the CNVs. We used a pool of other samples with CNVs different from the tested sequence as reference set. Default values of transition probability and the over-dispersion parameter of the binomial model were used in the analysis. 2.7. Statistical analysis Comparison of expected versus observed CNVs in our samples were done by goodness of fit test using 500.000 Monte Carlo simulations in R (www.r-project.org). For expected frequencies, we considered the frequencies based on the largest sample size between the studies of Rees et al. [6], Kirov et al. [4], or Malhotra and Sebat [3] (Table 1). In case of C16orf72 duplication, not reported on these studies, we used data from Levinson et al. [2]. CNVs for consideration were the C16orf72 duplication, the 17p12 deletion, associated with schizophrenia in the large study of Rees et al. [6], and those with penetrance for neuropsychiatric disorders more than zero according to Kirov et al. [4] with the exception of 22q11.2 duplication, that was removed from the analysis due to its putative protective role against schizophrenia [15]. VIPR2 was not included due to inconsistent results in bibliography related to its association with schizophrenia [6,12].

all for batch I, and were excluded from the analysis. An example of the distribution of relCN values is shown in Fig. 2, and all the distributions are present at Supplementary Fig. 2. As shown in these figures, although each individual assay may be informative for the identification of CNVs, the joint analysis of all the assays at each region presents a more consistent result. A total of 124 samples, presenting relCN N 2SD, were subjected to a second round of the method. A total of 8 CNVs in 14 samples were identified after two rounds of qicPCR (Table 2 and Supplementary Fig. 2). In four instances, the identified CNVs, 17p12 duplication, 15q13.3 duplication (twice) and 16p13.11 deletion, were the reciprocal of the known schizophrenia risk CNV [6]. In addition, one sample carried the VIPR2 duplication, whose involvement in psychiatric risk is unclear [6,12]. 3.3. Confirmation of CNVs MLPA data on CNVs at 15q13.3 or 16p11.2 confirmed the carrier status of the six samples identified by qicPCR (Table 2). There were no additional samples carrying these CNVs. Nine samples, presenting high risk CNVs according to qicPCR, were subject to exome sequencing for confirmation of the CNV. In any particular patient, 79.10% of bases were covered by at least 10× (average 86.09%), and 59.43% by at least 30 × (average 69.36%). This analysis confirmed the presence of the CNVs in the nine samples based on depth of coverage (Table 2, Fig. 3 and Supplementary Fig. 3). No false positive or negative findings were detected by the method. 3.4. Comparison with expected frequencies Assuming the population frequencies from large published datasets (Table 1), we detected a significant excess of CNVs associated with neuropsychiatric disorders in our sample (P goodness of fit = 0.00088). The 16p11.2 duplication was also highly significant (P goodness of fit = 0.000019 after Bonferroni's correction for 15 tests).

3. Results 4. Discussion 3.1. Design of the qicPCR assays First, we selected several positions at each CNV region differing between human and chimpanzee and located, if possible, in exons of genes at both edges of the CNV region and around the middle. Then, we used these positions as inputs for the Sequenom Assay Design software, altering the priority of inclusion of positions until three conditions were met: i) a multiplex assay has to include 25–30 assays; ii) all the assays from the same CNV region have to be included in the same multiplex assay; and iii) each CNV region has to be tested by a minimum of five assays. Different input sequences were tested until all three conditions were met. Two plexes were designed by this method. The final design is shown in Table 1. Multiplex 1 includes 25 assays and multiplex 2 includes 30 assays. Supplementary Table 1 shows the primers for each multiplex and Supplementary Fig. 1 shows an example of MassArray spectrum at one assay. 3.2. Identification of CNV carriers by qicPCR The overall design of the experiment is shown in Fig. 1. The method was tested in two batches of different sample sizes, 51 and 463, for batches I and II, respectively. Three assays (16p11MAZ, VIPR2_3 and VIPR2_7) of Multiplex 1, and 6 assays (16p13_11NDE1_8, 17q12DHRS11_1, 22q1121SCARF2, 22q1121THAP7_1, C16orf72_1, and C16orf72_1utr) of Multiplex 2, from batch I failed to amplify the human and/or chimpanzee variants in most of the samples and were excluded from the analysis. The same assays failed in batch II. Five samples presented missing data for more than 5% of the assays,

Here, we present a flexible, cheap and accurate method for screening of large collections of samples for the presence of CNVs of relevance in psychiatric genetics. The fundamentals of the method, qicPCR linked to variant detection by Sequenom MassArray technology, were first reported by Williams et al. [10]. The method has been used as a confirmation method by Vacic et al. [12] or Conrad et al. [16]. Nevertheless, this is the first time it is used to screen for the presence of multiple CNVs in a Table 2 Samples with CNVs discovered in this study. CNV

OR Schizophrenia (95% CI)a

Sample

MLPA confirmation

WES confirmation

del 15q11.2

2.15 (1.71–2.68)

dup 15q11.2 dup 15q13.3

13.20 (3.72–46.77)b n.a.

dup 16p11.2

11.52 (6.86–19.34)

del 16p13.11 dup 16p13.11

n.a. 2.30 (1.57–3.36)

dup 17p12 dup VIPR2

n.a. 1.54 (0.77–3.09)

SZ278 SZ459 SZ229c SZ374 SZ414 SZ014 SZ333 SZ347 SZ470 SZ429 SZ224 SZ541 SZ189 SZ338

n.a. n.a. Yes Yes Yes Yes Yes Yes Yes n.a. n.a. n.a. n.a. n.a.

Yes Yes Yes n.a. n.a. Yes Yes Yes Yes n.a. Yes Yes n.a. n.a.

a b c

Data from Rees et al. [6]. Data corresponding to the 15q11.2–13.1 duplication. MLPA confirmed that this CNV extends to region 15q13.1, not covered by qicPCR.

J. Rodriguez-Lopez et al. / Clinica Chimica Acta 445 (2015) 34–40

39

Fig. 3. CNV duplication in the 16p11.2 region of SZ333 sample identified by ExomeDepth in the exome sequence data. The gray area shows the estimated 99% confidence interval for the ratio of observed/expected number of reads taking into account the values of the reference set (absence of CNV). The black crosses show this value in each exon for the test sample.

large set of samples, validating the method as a high-throughput screening assay. In addition to the considerable increase in tested CNVs and samples, two main differences with previous uses should be stressed. First, we used five assays per CNV to improve accuracy as it has been shown previously that the accuracy of the method increased considerably if at least four assays were used [10]. Our data confirmed this observation. As shown in Fig. 2 and Supplementary Fig. 2, while the behavior of individual assays at the same CNV is similar, the use of several assays and analysis of the mean values per region increases the probability of detection of the CNVs. Second, we did not design assays at control regions. Instead, the assays at the other CNV regions were used as controls for each tested CNV. By this way, it is possible to accommodate more assays per CNV region in a single multiplex assay, increasing detection accuracy. The lack of control regions may be problematic for the detection of CNVs at samples with more than one of these CNVs, although taking into account their very low frequency and very low persistence in populations [5], the probability that an individual presents two different CNVs is extraordinarily low. One way to avoid this putative problem is to reanalyze data removing one CNV region each time. Williams et al. [10] suggested that the degree of divergence between humans and chimpanzees would allow for the use of this method to target a large majority of the human genome. Here, we show that in general this assumption is correct. However, three of the five assays for detection of CNVs at VIPR2 failed to amplify. This may be related to the fact that the selection of non-conserved nucleotide positions for design is limited to a single gene, reducing the possibility of combinations in multiplex assay designs. Nevertheless, the involvement of VIPR2 in schizophrenia risk is in doubt, after additional data in large samples showing no association [6]. One of the advantages of MassArray technology is its high flexibility, easily accommodating different assays to replace the less interesting CNVs. For instance, the 2q13 region or the 16p11.2 distal region may be an alternative to VIPR2 in new versions of the multiplex assay [8,17]. The main limit of the present work is that our sample size precludes an accurate estimation of sensitivity and specificity of the method. Larger samples sizes are needed to estimate these values. However, we have

detected a strong overrepresentation of CNVs in our schizophrenia sample, and the CNV carrier rate among our samples was similar to previous data [6,8], probing the validity of the method to detect association of this set of CNVs with psychiatric disorders. Genomic microarrays for detection of CNVs are the recommended first-tier diagnostic test for intellectual disabilities, multiple congenital anomalies, or autism spectrum disorders, rendering a 15–20% diagnostic yield [18]. The rate of carriers is expected to be considerably lower in schizophrenia, although a study on prospectively recruited community-based sample of 459 unrelated schizophrenia patients identified 8.1% of putatively clinically significant CNVs, opening the debate about the convenience of genomic microarray testing in schizophrenia [8]. Meanwhile, the existence of a cheap and accurate screening method for detection of the well-defined pathogenic CNVs may be an alternative option, as proposed earlier [19], taking into account its relevance in the context of genetic counseling and management of carriers [20–22]. In conclusion, we present a cheap, flexible, and accurate method for simultaneous screening of recurrent CNVs associated with neuropsychiatric disorders in large samples. Our method may be very useful in applications such as identification of carriers of these CNVs to characterize phenotypes affected by the CNVs related to physical health of psychiatric patients, such as nephropathies, obesity, hypotonia, or cardiac problems [21–23] as well as screening of samples of different neuropsychiatric disorders to fully delineate the range of clinical variability associated with each one of these CNVs. To this goal, we are currently applying the new version of the assays to screen collections of different psychiatric disorders. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.cca.2015.03.013.

Acknowledgments This work was supported by grant CP11/00163 from Instituto de Salud Carlos III/FEDER to JC. Genotyping was performed at the Santiago de Compostela node of Centro Nacional de Genotipado. We thank María Torres and Juan Ansede for their technical assistance.

40

J. Rodriguez-Lopez et al. / Clinica Chimica Acta 445 (2015) 34–40

References [1] Stefansson H, Rujescu D, Cichon S, et al. Large recurrent microdeletions associated with schizophrenia. Nature 2008;455:232–6. [2] Levinson DF, Duan J, Oh S, et al. Copy number variants in schizophrenia: confirmation of five previous findings and new evidence for 3q29 microdeletions and VIPR2 duplications. Am J Psychiatry 2011;168:302–16. [3] Malhotra D, Sebat J. CNVs: harbingers of a rare variant revolution in psychiatric genetics. Cell 2012;148:1223–41. [4] Kirov G, Rees E, Walters JTR, et al. The penetrance of copy number variations for schizophrenia and developmental delay. Biol Psychiatry 2013;75:378–85. [5] Rees E, Moskvina V, Owen MJ, O'Donovan MC, Kirov G. De novo rates and selection of schizophrenia-associated copy number variants. Biol Psychiatry 2011;70: 1109–14. [6] Rees E, Walters JTR, Georgieva L, et al. Analysis of copy number variations at 15 schizophrenia-associated loci. Br J Psychiatry 2014;204:108–14. [7] Girirajan S, Campbell CD, Eichler EE. Human copy number variation and complex genetic disease. Annu Rev Genet 2011;45:203–26. [8] Costain G, Lionel AC, Merico D, et al. Pathogenic rare copy number variants in community-based schizophrenia suggest a potential role for clinical microarrays. Hum Mol Genet 2013;22:4485–501. [9] Aradhya S, Cherry AM. Counting chromosomes to exons: advances in copy number detection. Curr Genet Med Rep 2013;1:71–80. [10] Williams NM, Williams H, Majounie E, et al. Analysis of copy number variation using quantitative interspecies competitive PCR. Nucleic Acids Res 2008;36:e112. [11] Carrera N, Arrojo M, Sanjuán J, et al. Association study of nonsynonymous single nucleotide polymorphisms in schizophrenia. Biol Psychiatry 2012;71:169–77. [12] Vacic V, McCarthy S, Malhotra D, et al. Duplications of the neuropeptide receptor gene VIPR2 confer significant risk for schizophrenia. Nature 2011;471:499–503.

[13] McKenna A, Hanna M, Banks E, et al. The genome analysis toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 2010;20:1297–303. [14] Plagnol V, Curtis J, Epstein M, 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. [15] Rees E, Kirov G, Sanders A, et al. Evidence that duplications of 22q11.2 protect against schizophrenia. Mol Psychiatry 2014;19:37–40. [16] Conrad DF, Pinto D, Redon R, et al. Origins and functional impact of copy number variation in the human genome. Nature 2010;464:704–12. [17] Guha S, Rees E, Darvasi A, et al. Implication of a rare deletion at distal 16p11.2 in schizophrenia. JAMA, Psychiatry 2013;70:253–60. [18] Miller DT, Adam MP, Aradhya S, et al. Consensus statement: chromosomal microarray is a first-tier clinical diagnostic test for individuals with developmental disabilities or congenital anomalies. Am J Hum Genet 2010;86:749–64. [19] Van Den Bossche MJ, Johnstone M, Strazisar M, et al. Rare copy number variants in neuropsychiatric disorders: specific phenotype or not? Am J Med Genet B Neuropsychiatr Genet 2012;159B:812–22. [20] Gershon ES, Alliey-Rodriguez N. New ethical issues for genetic counseling in common mental disorders. Am J Psychiatry 2013;170:968–76. [21] Tropeano M, Andrieux J, Collier DA. Clinical utility gene card for: 16p13.11 microdeletion syndrome. Eur J Hum Genet 2014;22:1–4. [22] Tropeano M, Andrieux J, Vassos E, Collier DA. Clinical utility gene card for: 15q13.3 microdeletion syndrome. Eur J Hum Genet 2014. http://dx.doi.org/10.1038/ejhg. 2014.88. [23] Girirajan S, Rosenfeld JA, Coe BP, et al. Phenotypic heterogeneity of genomic disorders and rare copy-number variants. N Engl J Med 2012;367:1321–31.