· Web viewM.I.N.I. alone cannot make a lifetime diagnosis of major (unipolar) depression and BDII, because it does not include questions to screen past major depressive episodes.

SUPPLEMENTARY INFORMATION

Exome Sequencing for Bipolar Disorder Points to Roles of De Novo Loss-of-function and Protein-altering MutationsRunning title: Exome sequencing points to de novo mutations in BD

Muneko Kataoka, MD1,2,6, Nana Matoba, MS1,3,6, Tomoyo Sawada, PhD1, An-a Kazuno, MS1, Mizuho Ishiwata1, Kumiko Fujii, MD, PhD1, 4, Koji Matsuo, MD, PhD5, Atsushi Takata, MD, PhD1,7

and Tadafumi Kato, MD, PhD 1,7

Author affiliation:1Laboratory for Molecular Dynamics of Mental Disorders, RIKEN Brain Science Institute, Saitama, 351-0198, Japan2Department of Child Neuropsychiatry, Graduate School of Medicine, The University of Tokyo, Tokyo, 113-8655, Japan3Department of Computational Biology and Medical Sciences, Graduate School of Frontier Sciences, The University of Tokyo, Chiba, 277-8561, Japan4Department of Psychiatry, Dokkyo Medical University School of Medicine, Tochigi, 321-0193, Japan5Division of Neuropsychiatry, Department of Neuroscience, Yamaguchi University Graduate School of Medicine, Yamaguchi, 755-8505, Japan6These authors contributed equally to this work and are listed in an alphabetical order7Co-corresponding authors

TABLE OF CONTENTSSUPPLEMENTARY MATERIALS AND METHODS..........................................................................3

Studied Subjects.......................................................................................................................... 3Library Preparation and Whole Exome Sequencing.....................................................................3Sequence Read Mapping and Variants Calling............................................................................4Identification of De Novo Point Mutations..................................................................................4Identification of De Novo Copy Number Variations.....................................................................5Data of De Novo Mutations in Controls and Patients with Schizoaffective Disorder...................6Enrichment Analysis of Loss-of-function and Protein-altering De Novo Mutations in Case Subjects....................................................................................................................................... 7Ages of Onset in Probands with or without De Novo Protein-Altering Mutations.......................7Gene Ontology Enrichment Analysis of Genes with Protein-Altering De Novo Mutations...........7Estimation of the Proportion of Genuine Disease-associated Mutations from Ascertainment Differentials.................................................................................................................................8Genes Hit by De Novo Protein-altering Mutations in BD and Also in Schizophrenia....................9Integrative Gene Ontology Enrichment Analysis of BD Candidate Genes....................................9Generation of Cells with the Frameshift Mutation in EHD1.........................................................9Immunoblot Analysis.................................................................................................................10

SUPPLEMENTARY FIGURES......................................................................................................11Figure S1. Sequencing Coverage in Each Individual and Each Trio.............................................11Figure S2. The De Novo 3q29 Deletion in BD.............................................................................12Figure S3. The De Novo Frameshift Mutation in EHD1 and Its Functional Consequence...........13

SUPPLEMENTARY TABLES........................................................................................................14Table S1. Detailed Information for the Studied Subjects and Sequencing Performance...........14Table S2. List of De Novo Mutations with Detailed Annotations...............................................15

SUPPLEMENTARY REFERENCES...............................................................................................17

2

SUPPLEMENTARY MATERIALS AND METHODSStudied Subjects

Participants were recruited through the Bipolar Disorder Research Network Japan or at

Yamaguchi University Hospital. All the probands are diagnosed with bipolar I or II disorder (BDI

or BDII) based on the DSM (Diagnostic and Statistical Manual of Mental Disorders) IV criteria by

trained psychiatrists. All the parents were screened for mental disorders by structured interview

using M.I.N.I. (Mini International Neuropsychiatric Interview)1. M.I.N.I. alone cannot make a

lifetime diagnosis of major (unipolar) depression and BDII, because it does not include questions

to screen past major depressive episodes. Therefore, additional questions to verify past history

of major depressive episodes were asked to the participants. We used DNA samples from 79

proband-parents trios in this study. We did not include families with a parent affected by bipolar

disorder (BD; BDI or BDII), schizophrenia or schizoaffective disorder. Because unipolar depression

is highly prevalent2, we did not exclude families with a history of unipolar depression. The

studied probands consisted of 32 males and 47 females, and 56 and 23 probands were

diagnosed with BDI and BDII, respectively. The average age at recruitment was 36.9 ± 9.2 (16-56)

years old. All the participants gave written informed consent for the study. The study was

approved by the First Committee of Research Ethics of RIKEN Wako Institute.

Library Preparation and Whole Exome Sequencing

Genomic DNA samples were obtained from either peripheral blood or saliva (Table S1). To

extract DNA from saliva, we used the Oragene kit (DNA Genotek Inc., Ontario, Canada)

according to the manufacturer’s instruction. Target capture of exome regions was performed on

individual samples using SureSelectXT Human All Exon kits V4, V5 or V5 + mitochondria (Agilent

Technologies, Inc., Santa Clara, CA). Whole exome sequencing (WES) was performed by using

either the HiSeq2000 or HiSeq2500 (Illumina, San Diego, CA) with paired-end 101bp reads.

Barcoded 5 or 6 libraries were pooled and sequenced in one lane of the HiSeq. Fastq files were

produced by the Illumina CASAVA pipeline. Raw sequence data will be available in dbGaP

(www.ncbi.nlm.nih.gov/gap).

3

Sequence Read Mapping and Variants Calling

We first excluded all the low quality reads (more than 20 % of bases failed Q20) from the fastq

files using the FASTQ Quality Filter module of FASTX-Toolkits-0.0.13.2 followed by compfast_pe.

(http://compbio.brc.iop.kcl.ac.uk/software/cmpfastq_pe.php). Then reads were mapped to the

Human 1kg Reference (GRCh37 + decoy) by using BWA-MEM3 (version 0.7.5a). Generated SAM

(sequence alignment/map) files were converted into the BAM (binary alignment/map) format

using SAMtools4. PCR duplicates were flagged by Picard (version 1.92,

http://picard.sourceforge.net/). Reads around known or private (family specific)

insertion/deletion (indel) sites were re-aligned locally using IndelRealigner module of the GATK 5

(version 2.6-4). GATK BaseRecalibrator was used to recalibrate base quality score. Single

nucleotide variants (SNVs) and indels were called by GATK UnifiedGenotyper from combined

bams of each trio. The standard parameters were used for hard filtering according to the GATK

best practices recommendations6. Only the positions with ≥ 20X coverage in all three family

members were considered for detection of de novo and transmittable variants.

Identification of De Novo Point Mutations

From the list of variant calls generated as above, we considered those with 1) eight or more

reads for the variant allele in the proband and 2) 95% or more reads for the reference allele in

both of the parents as candidates for de novo mutations except that they are within the HLA

locus. From the list of candidates for de novo mutations, we removed those found with a minor

allele frequency (MAF) greater than 0.01 in either of the following databases; 1) dbSNP

(http://www.ncbi.nlm.nih.gov/SNP/) build138 (except for those with clinical information), 2) the

NHLBI Exome Variant Server (http://evs.gs.washinton.edu.EVS/), 3) the 1000 Genomes Project

(http://www.1000genomes.org/) dataset and 4) the Exome Aggression Consortium (ExAC)

(http://exac.broadinstitute.org/) dataset. We also removed candidates that were found in

multiple families or found as a homozygous variant in one or more unaffected parents (likely

these variants are non-pathogenic or false positives). We finally performed manual inspection of

the candidates by using Integrative Genomics Viewer (IGV)7, and excluded four candidates that

are likely due to contamination of bacterial genomes according to the results of BLAST

4

(http://blast.ncbi.nlm.nih.gov/Blast.cgi) search. After performing these procedures, we

identified 78 candidates for de novo mutations. These candidates were verified by Sanger

sequencing with a standard protocol, and 91.0 % of them (71/78) were validated. We annotated

these validated de novo mutations by ANNOVAR (2015Mar22)8 with Ensembl Genes for

transcript mapping. If multiple annotations were assigned for one mutation, we used the

annotation with the most severe impact on protein (i.e., nonsense, canonical splice site,

frameshift indels > missense, inframe indels > synonymous). Based on these annotations we

classified the mutations into loss-of-function (LOF) mutations (nonsense, canonical splice site,

frameshift indels mutations), protein-altering mutations (LOF, missense and inframe indel

mutations) and synonymous mutations.

Identification of De Novo Copy Number Variations

We called copy number variations (CNVs) from our WES data using eXome Hidden-Markov

Model (XHMM)9-11 and Copy Number Inference From Exome Reads (CoNIFER)12. We only

subjected the genomic regions overlapped between the targets for SureSelect V4 and V5 to

these analyses. For the XHMM analysis, we first counted local read depth by using GATK

DepthOfCoverage function with the BAM files generated as above. Then the mean coverage of

the target regions for all exomes was merged into a matrix file. GC-rich regions, repeat

sequences and regions with outlier read-depth were excluded from the analysis. CNVs were

called by XHMM discover and genotype function with default parameters. For the CoNIFER

analysis, we first generated RPKM files from the BAM files and then called CNVs based on SVD-

ZRPKM values. From both of the datasets generated by the two software we excluded CNV calls

identified twice or more in our unaffected parental population and those not supported by the

coverage data of three or more genomic regions targeted by exome capture probes using

PLINK13. Then candidates for de novo CNVs were identified using PLINK/SEQ

(http://atgu.mgh.harvard.edu/plinkseq/). Among these candidates we subjected those called by

both software to validation experiments using the Human Genome CGH Microarrays (Agilent),

according to the results of a recent study demonstrating that most of the validated CNVs were

5

called by both software14.

Data of De Novo Mutations in Controls and Patients with Schizoaffective Disorder

We used the data of de novo mutations (SNVs and indels) in controls and patients with

schizoaffective disorder (SAD) in the following studies for comparison with our own data of BD:

controls; Iossifov et al.15 (# of trios =1,911), schizoaffective disorder; Xu et al.16 and McCarthy et

al.17 (# of trios = 64). We are aware that several previous studies suggested that schizoaffective

disorder, bipolar, is genetically more related to BD than schizoaffective disorder, depressed.

Thus, it would be better to include schizoaffective disorder, bipolar, only. However, in these

papers, no information of the subtype of schizoaffective disorder is listed. Several family studies,

however, support that depressive subtype of schizoaffective disorder is also associated with

elevated familial risk of BD18, 19. Therefore, we included both subtypes of schizoaffective disorder

to combine with BD. To compare the datasets from different studies in an equivalent condition,

we re-annotated and filtered all the de novo mutations from published studies with the same

pipeline used for our own data based on the genomic positions and the non-reference alleles of

mutations. It might be noteworthy that the rate of de novo LOF mutations among the control

subjects in Iossifov et al.15 could be slightly inflated because in their study candidates for de novo

LOF mutations were subjected to validation experiments more aggressively than others (for LOF

mutations they subjected candidates with the “strong” and “weak” tags for validation, while

they only considered the ones with the “strong” tag for other types of mutations).

Enrichment Analysis of Loss-of-function and Protein-altering De Novo Mutations in Case Subjects

To test enrichment of loss-of-function and protein-altering de novo mutations in BD, we

performed one-tailed Fisher’s exact tests with the following 2×2 table: columns; BD and controls

(1,911 unaffected siblings in Iossifov et al.15 as described above), rows; the number of de novo

LOF or protein-altering mutations and the number of synonymous mutations. This method

should be resistant to potential artifacts caused by comparison of data from different studies

(e.g. mutation detection rates may vary across studies) because synonymous mutations, whose

6

enrichment was not observed in a comparison between ASD and controls15, were used as the

internal control. Comparison between BD+SAD or BDI+SAD and controls was performed by using

the same procedures.

Ages of Onset in Probands with or without De Novo Protein-Altering Mutations

We compared average ages of onset between probands with one or more de novo protein-

altering mutations and those without mutations using two-tailed Student’s t-test. We also

analyzed averages ages of ascertainment between these two groups using two-tailed Student’s t-

test, to test whether significant difference in ages of onset can be explained by this factor.

Gene Ontology Enrichment Analysis of Genes with Protein-Altering De Novo Mutations

We performed a gene ontology (GO) enrichment analysis of genes hit by de novo protein-

altering mutations in BDI and SAD using Database for Annotation, Visualization and Integrated

Discovery (DAVID, v6.7)20, 21. We did not include the genes disrupted by the de novo CNV

identified in our cohort in the list of input genes because we cannot equally weight the CNV

disrupting multiple genes and a de novo mutation affecting a single gene. It is known that target

capture efficiency varies for different genes in exome sequencing22. This capture bias potentially

affects the results of GO enrichment analyses. We therefore used a custom background gene list

that does not include genes with poor coverage for DAVID analyses. To prepare the list, we first

defined “poor-coverage” regions using DiagnoseTargets module of GATK23 with default

parameters except for the minimum coverage set as 20 (same to the threshold used for variant

calling). We then calculated the proportion of exonic regions that overlap with the “poor-

coverage” regions for each gene using BEDTools24, and generated the custom list of background

genes by excluding the genes of which > 30% of the exonic regions were included in the “poor-

coverage” regions. Among the 17,392 genes assigned for any of the GO terms, 226 genes were

excluded by this procedure. We excluded extremely large (including > 5,000 genes) and small GO

terms (hit count less than five) from the result because these terms are in general less

informative. For the nine GO terms with nominal significance (P < 0.05) in this GO enrichment

analysis using DAVID, we further tested their specificity by performing a simulation analyses

7

randomly selecting 75 de novo protein-altering mutations (equal to the number of mutations in

BDI and SAD) in controls15 10,000 times. We considered that the GO terms that are not

significantly enriched in this simulation analysis do not represent the terms truly enriched

among the genes hit by protein-altering mutations in BDI and SAD, rather the enrichment of

these terms in our initial DAVID analysis can be explained by general properties of genes

preferentially hit by de novo mutations.

Estimation of the Proportion of Genuine Disease-associated Mutations from Ascertainment Differentials

Using our data of de novo mutations in BD and the data for controls in Iossifov et al.15, we

calculated per-individual mutation rates as follows; per-individual rates for de novo LOF

mutations, 0.114 in BD and 0.089 in controls, per-individual rates for de novo protein-altering

mutations, 0.722 in BD and 0.657 in controls. By using these numbers and the procedures

described in Iossifov et al.15, we calculated ascertainment differentials as (per-individual rates of

de novo LOF or protein-altering mutations in our BD cohort) - (per-individual rates of de novo

LOF or protein-altering mutations in controls). Based on these ascertainment differentials we

roughly estimated the proportion of the mutations contributing to the disease risks as follows:

LOF mutations; (0.114 - 0.089) / 0.114 = 0.22, protein-altering; (0.722 - 0.657) / 0.722 = 0.09.

Genes Hit by De Novo Protein-altering Mutations in BD and Also in Schizophrenia

The list of de novo mutations in schizophrenia was obtained from Refs 16, 17, 25-28. From this list we

identified five genes that are hit by de novo protein-altering mutations in BD and also in

schizophrenia. An expected number of de novo protein-altering mutations for each gene was

calculated from per-gene mutation rates provided in Supplementary Table 1 of Samocha et al.29

Then we invoked Poisson distribution probabilities to determine the significance for the

observed number of mutations (note that these P values are not subjected to genome-wide

correction for multiple testing).

8

Integrative Gene Ontology Enrichment Analysis of BD Candidate Genes

To perform an integrative gene ontology enrichment analysis of BD candidate genes we

included the following gene lists in the input; 1) genes with de novo protein-altering mutations

in our WES study (# = 56), 2) genes with SNPs associated with BD at P < 1 × 10 -4 in a large-scale

GWAS30, (# = 54) and 3) genes included in CNVs that showed nominally significant association

with BD (Table 1 and Supplementary Table S4 of Green et al.31, # = 120). By using these input

genes (# = 229 after excluding overlaps) we performed an enrichment analysis using DAVID.

Generation of Cells with the Frameshift Mutation in EHD1

HEK293T (RIKEN Cell Bank) cells were maintained at 37°C with 5% CO2 atmosphere in DMEM

(Wako Pure Chemical Industries, Ltd., Osaka, Japan) supplemented with 10% FBS (Life

Technologies Japan, Tokyo, Japan). Plasmids were transfected using Lipofectamine 2000 (Life

Technologies Japan) according to manufacturer’s instructions. Human cDNA for EHD1 was

purchased from Kazusa DNA Res. Inst. and cloned into pcDNA3-Myc vector. The 1414 delG

mutation was introduced into hEHD1 by PCR-based mutagenesis using 5’-

TGAAGTCCAAGCTCCCCAAC-3’ and 5’- ATCTCCTTCTTGGCGTTGG-3’ as primers. The full-length

sequence of hEHD1 1414delG was confirmed by Sanger sequencing.

Immunoblot Analysis

Cells were harvested 28 hrs post-transfection and lysed in 1% Triton X-100 -based lysis buffer

(10 mM Tris-HCl [pH 7.4], 120 mM NaCl, 5 mM EDTA, 1% Triton X-100 and protease inhibitor

[Roche Diagnostics, Tokyo Japan]). Cell lysates were subjected to Western blot analysis with

detection reagents (Termo Fisher Scientific, Waltham, MA, USA). Antibodies used in this study

are as follows: anti-Myc (4A6) (Merck Millipore, Billerica, MA, USA), anti-EHD1 (ab75886 and

ab109747) (Abcam, Cambridge, UK), anti-β-actin (AC-15) (Sigma-Aldrich Japan, Tokyo, Japan),

goat anti-Mouse IgG-HRP and goat anti-Rabbit IgG-HRP (SantaCruz Biotechnology, Santa Cruz,

CA, USA).

9

SUPPLEMENTARY FIGURES

Figure S1. Sequencing Coverage in Each Individual and Each Trio

Proportion of the exome target regions with joint coverage (the coverage of the least well

covered individual in the trio) ≥ 20 was plotted as red circles in the order of the trio rank (trios

with the lowest coverage on the left and the highest on the right). Proportion of the target

regions covered by ≥ 20 reads at the individual level (blue open circles) and on average in the

trio (blue circles) was also plotted.

10

Figure S2. The De Novo 3q29 Deletion in BD

(A) Visualization of the output from CoNIFER12 for the 3q29 region. Red, blue and green lines

indicate normalized coverage in the proband, father and mother, respectively. (B) Signals from

array CGH visualized by using the Agilent Genomic Workbench. Reduction of signals only in the

proband indicating existence of a de novo deletion was detected in the same region in (A) and

(B). (C) A cytoband image for the human chromosome 3. A red line indicates the locus where the

deletion was detected.

11

Figure S3. The De Novo Frameshift Mutation in EHD1 and Its Functional Consequence

(A) Electropherogram of Sanger sequencing of EHD1. The heterozygous 1414G deletion in EHD1

was detected only in the proband. (B) Schematic representation of wild-type and 1414delG

mutant EHD1 protein. The de novo frameshift mutation directory introducing a stop codon

results in truncation of the C-terminal of the protein. (C) Western blot analysis showing the

expression of wild-type and 1414delG mutant EHD1. Although anti-Myc antibody or anti-EHD1

antibody against a.a.390-415 detected both wild-type and mutant EHD1, mutant EHD1 was not

detected by anti-EHD1 antibody against a.a.500-534, suggesting expression of the truncated

protein.

12

SUPPLEMENTARY TABLES

Table S1. Detailed Information for the Studied Subjects and Sequencing PerformanceThis table is provided as a separate spreadsheet

13

Table S2. List of De Novo Mutations with Detailed AnnotationsThis table is provided as a separate spreadsheet

14

Table S3. List of 75 Genes Subjected to Gene Ontology Enrichment Analyses

15

SUPPLEMENTARY REFERENCES

1. Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E et al. The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J Clin Psychiatr 1998; 59 Suppl 2: 22-33.

2. Kessler RC, Berglund P, Demler O, Jin R, Koretz D, Merikangas KR et al. The epidemiology of major depressive disorder: results from the National Comorbidity Survey Replication (NCS-R). JAMA 2003; 289: 3095-3105.

3. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 2009; 25: 1754-1760.

4. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 2009; 25: 2078-2079.

5. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 2010; 20: 1297-1303.

6. DePristo Ma, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet 2011; 43: 491-498.

7. Robinson JT, Thorvaldsdottir H, Winckler W, Guttman M, Lander ES, Getz G et al. Integrative genomics viewer. Nature biotechnology 2011; 29: 24-26.

8. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic acids research 2010; 38: e164.

9. Fromer M, Moran JL, Chambert K, Banks E, Bergen SE, Ruderfer DM et al. Discovery and statistical genotyping of copy-number variation from whole-exome sequencing depth. Am J Hum Genet 2012; 91: 597-607.

10. Fromer M, Purcell SM. Using XHMM Software to Detect Copy Number Variation in Whole-Exome Sequencing Data. Current Protocols in Human Genetics 2014; 81: 7.23.21-

16

27.23.21.

11. Poultney CS, Goldberg AP, Drapeau E, Kou Y, Harony-Nicolas H, Kajiwara Y et al. Identification of small exonic CNV from whole-exome sequence data and application to autism spectrum disorder. Am J Hum Genet 2013; 93: 607-619.

12. Krumm N, Sudmant P, Ko A. Copy number variation detection and genotyping from exome sequence data. Genome Res 2012; 22: 1525-1532.

13. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. American journal of human genetics 2007; 81: 559-575.

14. Krumm N, Turner TN, Baker C, Vives L, Mohajeri K, Witherspoon K et al. Excess of rare, inherited truncating mutations in autism. Nat Genet 2015; 47: 582-588.

15. Iossifov I, O’Roak BJ, Sanders SJ, Ronemus M, Krumm N, Levy D et al. The contribution of de novo coding mutations to autism spectrum disorder. Nature 2014; 515: 216-221.

16. Xu B, Ionita-Laza I, Roos JL, Boone B, Woodrick S, Sun Y et al. De novo gene mutations highlight patterns of genetic and neural complexity in schizophrenia. Nat Genet 2012; 44: 1365-1369.

17. McCarthy SE, Gillis J, Kramer M, Lihm J, Yoon S, Berstein Y et al. De novo mutations in schizophrenia implicate chromatin remodeling and support a genetic overlap with autism and intellectual disability. Mol Psychiatry 2014; 19: 652-658.

18. Kendler KS, Gruenberg AM, Tsuang MT. A DSM-III family study of the nonschizophrenic psychotic disorders. The American journal of psychiatry 1986; 143: 1098-1105.

19. Kendler KS, McGuire M, Gruenberg AM, Walsh D. Examining the validity of DSM-III-R schizoaffective disorder and its putative subtypes in the Roscommon Family Study. The American journal of psychiatry 1995; 152: 755-764.

20. Huang DW, Sherman BT, Lempicki Ra. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 2009; 4: 44-57.

21. Huang DW, Sherman BT, Lempicki Ra. Bioinformatics enrichment tools: paths toward the

17

comprehensive functional analysis of large gene lists. Nucleic Acids Res 2009; 37: 1-13.

22. Meynert AM, Ansari M, FitzPatrick DR, Taylor MS. Variant detection sensitivity and biases in whole genome and exome sequencing. BMC Bioinformatics 2014; 15: 247.

23. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome research 2010; 20: 1297-1303.

24. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 2010; 26: 841-842.

25. Fromer M, Pocklington AJ, Kavanagh DH, Williams HJ, Dwyer S, Gormley P et al. De novo mutations in schizophrenia implicate synaptic networks. Nature 2014; 506: 179-184.

26. Girard SL, Gauthier J, Noreau A, Xiong L, Zhou S, Jouan L et al. Increased exonic de novo mutation rate in individuals with schizophrenia. Nat Genet 2011; 43: 860-863.

27. Gulsuner S, Walsh T, Watts AC, Lee MK, Thornton AM, Casadei S et al. Spatial and temporal mapping of de novo mutations in schizophrenia to a fetal prefrontal cortical network. Cell 2013; 154: 518-529.

28. Takata A, Xu B, Ionita-Laza I, Roos JL, Gogos JA, Karayiorgou M. Loss-of-function variants in schizophrenia risk and SETD1A as a candidate susceptibility gene. Neuron 2014; 82: 773-780.

29. Samocha KE, Robinson EB, Sanders SJ, Stevens C, Sabo A, McGrath LM et al. A framework for the interpretation of de novo mutation in human disease. Nat Genet 2014; 46: 944-950.

30. Sklar P, Smoller JW, Fan J, Ferreira MAR, Perlis RH, Chambert K et al. Whole-genome association study of bipolar disorder. Mol Psychiatry 2008; 13: 558-569.

31. Green EK, Rees E, Walters JTR, Smith K-G, Forty L, Grozeva D et al. Copy number variation in bipolar disorder. Mol Psychiatry 2016; 21: 89-93.

18