Skip to main content
  • Loading metrics

An interaction-based model for neuropsychiatric features of copy-number variants

  • Matthew Jensen,

    Affiliations Department of Biochemistry and Molecular Biology, Pennsylvania State University, University Park, Pennsylvania, United States of America, Bioinformatics and Genomics Program, Pennsylvania State University, University Park, Pennsylvania, United States of America

  • Santhosh Girirajan

    Affiliations Department of Biochemistry and Molecular Biology, Pennsylvania State University, University Park, Pennsylvania, United States of America, Bioinformatics and Genomics Program, Pennsylvania State University, University Park, Pennsylvania, United States of America, Department of Anthropology, Pennsylvania State University, University Park, Pennsylvania, United States of America


Variably expressive copy-number variants (CNVs) are characterized by extensive phenotypic heterogeneity of neuropsychiatric phenotypes. Approaches to identify single causative genes for these phenotypes within each CNV have not been successful. Here, we posit using multiple lines of evidence, including pathogenicity metrics, functional assays of model organisms, and gene expression data, that multiple genes within each CNV region are likely responsible for the observed phenotypes. We propose that candidate genes within each region likely interact with each other through shared pathways to modulate the individual gene phenotypes, emphasizing the genetic complexity of CNV-associated neuropsychiatric features.

A case for a multigenic model of CNV pathogenicity

Since the advent of large-scale sequencing studies, the number of genes associated with neurodevelopmental disorders such as autism, intellectual disability, and schizophrenia has increased dramatically. For example, nearly 200 genes have been identified with recurrent de novo mutations in both individuals with autism and intellectual disability [18]. In fact, complex human disease phenotypes can be influenced by variation in both a small number of core genes with large effect size and a large number of modifier genes with small effect size, accounting for the large number of candidate neurodevelopmental genes [9,10]. The application of a multigenic model for disease pathogenicity has not been fully expanded to cover copy-number variants (CNVs), or large duplications and deletions in the genome. The prevailing notion of single causative genes for CNV disorders is due to the paradigm of gene discoveries for CNVs associated with genetic syndromes in individuals with specific constellations of clinical features, such as Smith-Magenis syndrome (SMS). Although some variability in phenotypic expression has been documented, these disorders usually occur de novo and are characterized by high penetrance for the observed phenotypes [11,12] (Fig 1, S1 Table, S2 Table). In these cases, individuals manifesting the characteristic features of the syndrome but with either atypical breakpoints or mutations in individual genes within the CNV region were used to identify causative genes for the major phenotypes [1315]. These causative genes, such as RAI1 for SMS, were then confirmed by recapitulating conserved phenotypes of the deletion using functional evaluations in animal models [16,17].

Fig 1. Phenotypic profiles of syndromic and variably expressive CNVs.

(A) Table listing variably expressive (top) and syndromic (bottom) CNV regions is shown. The colored boxes indicate the frequency of de novo versus inherited CNV cases for del and dup previously identified in a cohort of 2,312 children with developmental disorders [12]. The 12 variably expressive CNV regions highlighted in bold were selected for the analysis described in the article. (B) Table listing average frequencies of neurodevelopmental phenotypes for select variably expressive and syndromic CNVs, curated from GeneReviews reports on individual CNVs [18], is shown. White boxes represent no available data from GeneReviews but do not necessarily indicate a lack of association between the CNV and the phenotype (for example, 1q21.1 deletion and schizophrenia). Data for this figure are available in S1 Table and S2 Table. CNV, copy-number variant; del, deletion; dup, duplication.

In contrast, another category of CNVs has been identified in individuals with neurodevelopmental disorders, including duplications and deletions at proximal 16p11.2, 3q29, distal 16p11.2, and 1q21.1 [1922]. Although these CNVs are enriched in affected individuals compared to population controls, they are primarily characterized by variable expressivity of clinical features [12,2327] (Fig 1B, S2 Table). For example, the 16p11.2 deletion has been implicated in 1% of individuals with idiopathic autism [19,28], but only 25% of individuals with the deletion exhibit an autism phenotype [2932], whereas others may manifest intellectual disability, obesity, or epilepsy at varying degrees of penetrance [29,33,34]. In fact, certain CNVs, such as the 16p12.1 deletion and the 15q11.2 deletion, have a high frequency of carriers who only manifest mild neuropsychiatric features, in contrast to more severely affected individuals who also carry other rare variants in the genetic background [12,23,24,27,35,36]. As such, many variably expressive CNVs have a higher frequency of inherited compared to de novo occurrence [12] (Fig 1A, S1 Table).

Based on the success of gene discovery in CNVs with syndromic features, such as SMS, several studies have attempted to identify the causative genes in variably expressive CNVs [3753]. Several individual genes within variably expressive CNV regions have been associated with specific congenital or structural features of these disorders, including TBX6 for scoliosis in 16p11.2 deletion [54], TBX1 for cardiac phenotypes in 22q11.2 deletion [39,55], GJA8 for cataracts and GJA5 for heart defects in 1q21.1 deletion [56,57], and MYH11 for aortic aneurysms in 16p13.11 duplication [58,59]. However, approaches to identify single causative genes for the more prominent neuropsychiatric features of these CNVs have not been successful [60]. Here, we show several lines of evidence from gene pathogenicity metrics, animal model studies, and gene expression data that support the involvement of multiple genes towards the neuropsychiatric features of variably expressive CNVs.

First, genome-wide metrics of pathogenicity, including those that measure haploinsufficiency (haploinsufficiency score [HI]; essentiality score; genome-wide haploinsufficiency score [GHIS], and EpiScore) [6164] and resistance to variation (residual variance to intolerance score [RVIS], probability of loss-of-function intolerance [pLI], and maximum constrained coding region [CCR] scores) [6567], provide evidence for several candidate genes within CNV regions for developmental disorders (Fig 2A, S3 Table). For example, 45 out of 152 genes (30%) within 12 variably expressive CNV regions are intolerant to variation with RVIS metrics in the top 20th genome-wide percentile, similar to that of known neurodevelopmental genes such as CHD8, NRXN1, and SCN2A, as well as genes responsible for major features of syndromic CNVs, such as RAI1 and NSD1 (S3 Table). These top-ranked genes include TAOK2, MVP, ALDOA, and DOC2A on chromosome 16p11.2, BCL9 and GJA5 on chromosome 1q21.1, and ATXN2L, ATP2A1, and SH2B1 on distal 16p11.2 (Fig 2A). Similarly, 32/165 genes (19%) are considered intolerant to loss-of-function mutations based on pLI scores (>0.9), and 36/160 genes (23%) have HI scores in the highest 20th percentile of the entire genome (S3 Table). Furthermore, the top 10% of all genes identified by a gene interaction-based machine-learning classifier to be associated with autism included 8 genes within 16p11.2 and 4 genes within 22q11.2 [68].

Fig 2. Pathogenicity metrics and model organism phenotypes for CNV genes.

(A) Percentile-rank scores compared to the whole genome for intolerance to variation (RVIS, pLI, and maximum CCR) and haploinsufficiency (HI, Essentiality, GHIS, and EpiScore) metrics for genes within select variably expressive CNV regions are shown [6167]. Lower percentile scores indicate a gene is more likely to be haploinsufficient or intolerant to variation. Gray boxes indicate that metrics were not available for a particular gene. (B) Developmental phenotypes in animal models for homologs of individual genes within the 16p11.2 region, as cataloged from animal model databases (MGI, ZFIN, and FlyBase), are shown. Black boxes indicate presence of phenotype, white boxes indicate absence of phenotype, and gray boxes indicate that no homolog is present for a particular gene in that model organism. The phenotypes observed in 16p11.2 deletion and duplication mice are distinct from those observed in the individual gene models [6973]. Data for this figure, including gene metrics and animal phenotypes for other CNV genes not shown in this figure, are available in S3 Table and S4 Table. CCR, constrained coding region; CNV, copy-number variant; GHIS, genome-wide haploinsufficiency score; HI, haploinsufficiency score; MGI, Mouse Genome Informatics; pLI, probability of loss-of-function intolerance; RVIS, residual variance to intolerance score; ZFIN, Zebrafish Information Network.

Second, several recent studies using animal and cellular models have demonstrated the critical involvement of several genes within CNVs towards neurological, cellular, and developmental functions [38,47,48,52,53,74] (Fig 2B, S4 Table). For example, Blaker-Lee and colleagues screened 22 homologs of 16p11.2 genes in zebrafish morpholino knockdown models and identified 20 homologs that contributed to morphological defects and abnormal behavior [38]. Iyer and colleagues also screened homologs of 16p11.2 genes in Drosophila melaogaster using RNA interference (RNAi) knockdown and found that 10 out of 14 homologs contributed to global developmental defects as well as specific neuronal and cellular defects in the developing fly eye [47]. Further, mouse models for 15 genes within the 16p11.2 region have been generated to test for defects in development and neuronal behavior [46,4951,7587]. For example, Taok2-/- mice have increased brain size, behavioral defects, and impaired synapse development [51]; Kcdt13+/- mice show defects in hippocampal synaptic transmission and decreased dendritic complexity [46]; Mapk3+/- mice show behavior anomalies, abnormal synapse function, and reduced cell proliferation during development [75,76]; and Mvp+/- mice show decreased plasticity and synaptic defects in ocular neurons [49] (Fig 2B). These models of individual genes do not fully recapitulate the phenotypes observed in models of the entire CNV [6973]. For example, the decreased body weight, abnormal brain morphology, and coordination defects observed in 16p11.2 deletion mouse models have not been observed in any individual gene knockdown models [6972] (Fig 2B). Similarly, Otud7a+/- mouse models have low body weight, reduced vocalization, abnormal dendritic spine morphology, and seizures, but the 15q13.3 deletion mice also show learning and memory defects in addition to the above features [44,45,88]. Furthermore, mouse models for Chrna7+/-, another candidate gene on chromosome 15q13.3, only show subtle behavioral phenotypes [89]. These data suggest that haploinsufficiency of CHRNA7 or OTUD7A alone is not sufficient to account for the pathogenicity of the entire CNV. Overall, a catalog of functional data from mouse [90], zebrafish [91], and fruit fly studies [92] indicates that 80% (131/163) of homologs for genes within CNV regions present lethality, behavioral, developmental, or neuronal phenotypes when disrupted (S4 Table). These data suggest that disruption of multiple genes within each CNV region can affect important developmental or neuronal functions that could contribute to the phenotypes of the entire CNV.

Third, patterns of gene expression in humans and model organisms have identified multiple genes within each CNV region that are co-expressed in the developing brain along with known neurodevelopmental genes. For example, Maynard and colleagues examined expression patterns of 22q11.2 gene homologs in the developing mouse brain and found that 27 out of 32 genes were expressed in the embryonic forebrain, with six genes expressed in neuronal tissues related to schizophrenia [40]. In fact, a genome-wide weighted gene correlation network analysis (WGCNA) [93] from different brain tissues during development [94] shows several large modules of genes with similar expression patterns (Fig 3, S5 Table). For example, the five largest modules are each enriched (p < 0.05 with Benjamini-Hochberg correction) for biological functions related to neurodevelopment, including protein modification and transport in module 1 (M1), nervous system development in M2, and cell communication and signal transduction in M5. Each of these modules contains multiple genes from the same CNV region, including 3q29 genes PAK2, NCBP2, and BDH1 in M1, 1q21.1 genes BCL9, CHD1L, and FMO5 in M2, and 16p11.2 genes MVP and QPRT in M5. Therefore, it is clear that multiple genes in the same CNV region are co-expressed with each other in the developing brain and could share similar functions or regulatory patterns.

Fig 3. CNV gene co-expression in the developing brain.

Modules of co-expressed genes derived from WGCNA analysis of BrainSpan Atlas RNA-Seq data (Gencode v. 10) [94] across 524 tissues and time points the developing brain are shown. Networks of interactions among genes within three select top WGCNA modules (M1, M2, and M5) were obtained from the BioGrid interaction database [95] and visualized using Cytoscape [96]. Genes within variably expressive CNV regions are highlighted as colored nodes in each network. Bar graphs show enrichment (p < 0.05 with Benjamini-Hochsberg correction, represented by red dotted line) of genes within each module for GO Biological Process terms, calculated using PantherDB [97]. Data for this figure are available in S5 Table. CNV, copy-number variant; GO, Gene Ontology; WGCNA, weighted gene correlation network analysis.

Dissecting the genetic complexity of CNV pathogenicity

Several scenarios could explain how the haploinsufficiency of multiple genes can predict the variable phenotypes associated with the entire CNV (Fig 4A). The simplest such model is an additive model, in which disruption of individual genes within a CNV may only impart a mild phenotype on their own but additively contribute to more severe features [37] (Fig 4A). However, an additive model may not always explain the phenotypic features manifested by CNVs containing multiple candidate genes that could lead to severe defects or lethality on their own. For example, heterozygous Tbx1+/- (within the 22q11.2 region) and Mapk1+/- (within the distal 22q11.2 region) mice both lead to perinatal or neonatal lethality [98100]. In humans, 14% (24/172) of CNV genes are under evolutionary constraint in control populations (pLI score > 0.9 or maximum CCR score greater than the 99th percentile) and have no reported disease-associated variants [101103], suggesting that these genes could be under strong purifying selection [67]. Furthermore, 18% (22/125) of CNV genes show evolutionary constraint for loss-of-function mutations (pLI > 0.9) but not for copy-number changes within a control population [104]. We therefore hypothesize that the pathogenicity of variably expressive CNVs can also be explained by complex interactions among the constituent genes within shared biological pathways. These interactions can enhance or suppress the phenotypes caused by disruption of individual genes. Under this model, the haploinsufficiency of certain genes can be modulated by haploinsufficiency of other interacting genes in the same region that may or may not lead to phenotypes on their own (Fig 4A). Furthermore, variants in the genetic background that map within these shared pathways can simultaneously modulate the effects of multiple genes, ultimately defining the phenotypic trajectory in CNV carriers (Fig 4A). For example, Pizzo and colleagues found that the burden of rare deleterious mutations within genes in the genetic background correlated with variability of IQ scores and head circumference among 16p11.2 deletion carriers [36]. The potential for complex interactions within a CNV region depends on the functional convergence of the constituent genes. For instance, both KCTD13 and TAOK2 within 16p11.2 participate in the Ras homolog A (RhoA) signaling pathway [46,51] and therefore are more likely to interact with each other than genes located in different biological pathways. In fact, it has been shown that genes within pathogenic CNVs are more similar in function compared to genes within benign CNVs, suggesting that variably expressive CNVs are likely to contain interactions between functionally relevant genes [105]. Further, Noh and colleagues found an overrepresentation of interactions among genes within autism-associated CNVs, and these interactions were enriched for synaptic transmission and regulatory signaling pathways [106]. Because of this, therapeutic targets for pathways shared among CNV genes could be explored as potential treatments for CNV disorders.

Fig 4. Models for genetic interactions within CNV regions.

(A) Several models of interactions among CNV genes are shown. These models include (i) a single-gene model in which one gene is sufficient to account for the phenotype; additive models in which the phenotype is due to the additive effects of multiple CNV genes that (ii) may or (iii) may not account for phenotypes on their own; and (iv) a complex interaction model in which additive, enhancer, and suppressor interactions between genes in the CNV region modulate the phenotype, including when additive effects of the CNV genes would lead to lethality on their own (gray circles). The size of the circles in the plot indicates the relative contribution of each gene to the overall neurodevelopmental phenotype. Thick circles indicate genes that contribute to the observed phenotypes on their own, and connector lines indicate the nature of interaction between pairs of genes. Connected modifier genes (M) can further modulate these interactions to ultimately define the phenotypic trajectory in individuals carrying the CNV. (B) For a hypothetical CNV region with three genes, there are seven combinations of gene knockdowns (A, B, C, AB, AC, BC, and ABC) that can be tested for the presence or absence of a specific phenotype. These knockdown experiments can yield 128 potential outcomes for each phenotype tested, with each individual set of outcomes corresponding to 1 of 64 combinations of pairwise gene interactions (additive, enhancer, suppressor, or no interaction). One possible outcome, highlighted in orange, shows presence of a particular phenotype for knockdowns of single genes A and B and two-hit knockdowns AB and BC. The single-gene knockdowns indicate that only genes A and B contribute to the phenotype and that the phenotype of pairwise knockdown AB is due to the additive effects of the two genes. Although the phenotype is observed for BC knockdown, the phenotype is not observed for AC and ABC knockdowns, suggesting that gene C suppresses the phenotype of gene A. CNV, copy-number variant.

The possibility of additive, suppressor, and enhancer interactions between pairs of genes underlies the potential for highly complex models of CNV pathogenicity. For instance, within a CNV region spanning three genes, seven combinations of gene knockdown experiments (haploinsufficiency of A, B, C, AB, BC, AC, and ABC) can be tested for the presence or absence of a specific phenotype (Fig 4B). This set of knockdown experiments can yield 128 possible experimental outcomes that can be used to further deduce 64 possible sets of pairwise interactions for AB, BC, and AC (no interaction, additive, suppression, or enhancement for each interaction) (Fig 4B). These possible combinations of interactions exponentially increase for larger CNVs with more genes, and the complexity further increases if quantitative phenotypes are used to determine the magnitude of interactions between genes or when interactions with variants in the genetic background are taken into account. However, testing even a small number of these interactions would still uncover the nature of the relationships among genes within a CNV region and potentially a common pathway shared by those genes. For example, Grice and colleagues used D. melanogaster RNAi models to identify 6 synergistic interactions out of 41 tested pairwise interactions between genes within de novo CNVs from autism patients, including partial 3q29 and 22q11.2 deletions [107]. Iyer and colleagues also used fly models to identify 24 additive, enhancer, and suppressor interactions out of 52 tested pairwise interactions among homologs of 16p11.2 genes [47], providing further evidence for complex interactions within CNV regions. Furthermore, these interaction models for CNV pathogenicity can be tested in cellular models of the entire CNV. For example, a more severe phenotype observed by restoring dosage of a candidate gene would suggest that disruption of this gene potentially suppresses the effects of other genes within the CNV.

Complex genetic interactions in the context of genome sequencing

In recent years, exome and whole-genome sequencing analysis has proven invaluable in identifying candidate genes for neurodevelopmental disorders [108]. However, sequencing studies would not be able to capture the genetic complexity of a multigenic CNV region. For example, genes that cause severe phenotypes or lethality on their own and are modulated by haploinsufficiency of other interacting genes within a CNV are less likely to have an enrichment of mutations in sequencing studies. Furthermore, because of the strong phenotypic heterogeneity of these CNVs, it is not possible to determine whether the phenotypes of any individual candidate gene fully recapitulate the variable phenotypes of the entire CNV region. Candidate genes within CNVs identified through genome sequencing studies, such as TAOK2 on chromosome 16p11.2 [51] or CHRNA7 on chromosome 15q13.3 [109], do not preclude the possibility of other candidate genes in the same region. Because of this, a thorough systems-based approach for each gene within a CNV and its interactions is necessary to identify candidate genes responsible for the neuropsychiatric features of each region [110].

In summary, genomic and functional data have implicated multiple genes in variably expressive CNV regions toward neuropsychiatric phenotypes, suggesting that single causative genes are not responsible for the heterogeneous features of these CNVs. Here, we propose a complex interaction-based model for these CNVs, in which candidate genes within each region interact with each other to influence the variable clinical outcome. The CNV phenotype is therefore distinct from the phenotype manifested by any individual gene, or in some cases, the additive effects of all genes in the region. This multigenic model of CNVs agrees with a broader complex genetic view of neurodevelopmental disorders, in which hundreds of genes with varying effect sizes and complex interactions influence developmental features [10]. Further studies on the role of individual genes in CNV regions towards neurodevelopment, especially those that identify key interactions between genes, will be useful in uncovering the cellular pathways and mechanisms responsible for the observed neuropsychiatric features.

Supporting information

S1 Table. De novo and inherited occurrence of syndromic and variably expressive CNVs.

CNV, copy-number variant.


S2 Table. Frequency of neuropsychiatric phenotypes among CNV carriers.

CNV, copy-number variant.


S3 Table. Metrics of intolerance to variation and haploinsufficiency for CNV genes (raw scores and percentiles).

CNV, copy-number variant.


S4 Table. Developmental phenotypes of CNV genes in mouse, zebrafish, and Drosophila models.

CNV, copy-number variant.


S5 Table. WGCNA modules for CNV genes and GO biological process enrichments of select modules.

CNV, copy-number variant; GO, Gene Ontology; WGCNA, weighted gene correlation network analysis.



The authors thank Lucilla Pizzo, Vijay Kumar, and Maitreya Das from the Girirajan Lab for their helpful discussions and comments on the manuscript.


  1. 1. O’Roak BJ, Vives L, Fu W, Egertson JD, Stanaway IB, Phelps IG, et al. Multiplex targeted sequencing identifies recurrently mutated genes in autism spectrum disorders. Science. 2012;338: 1619–1622. pmid:23160955
  2. 2. Sanders SJ, Murtha MT, Gupta AR, Murdoch JD, Raubeson MJ, Willsey AJ, et al. De novo mutations revealed by whole-exome sequencing are strongly associated with autism. Nature. 2012;485: 237–241. pmid:22495306
  3. 3. 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. pmid:25363768
  4. 4. Turner TN, Hormozdiari F, Duyzend MH, McClymont SA, Hook PW, Iossifov I, et al. Genome Sequencing of Autism-Affected Families Reveals Disruption of Putative Noncoding Regulatory DNA. Am J Hum Genet. 2016;98: 58–74. pmid:26749308
  5. 5. Yuen RKC, Merico D, Bookman M, Howe JL, Thiruvahindrapuram B, Patel R V, et al. Whole genome sequencing resource identifies 18 new candidate genes for autism spectrum disorder. Nat Neurosci. 2017;20: 602–611. pmid:28263302
  6. 6. Gilissen C, Hehir-Kwa JY, Thung DT, Van De Vorst M, Van Bon BWM, Willemsen MH, et al. Genome sequencing identifies major causes of severe intellectual disability. Nature. 2014;511: 344–347. pmid:24896178
  7. 7. Deciphering Developmental Disorders Study. Prevalence and architecture of de novo mutations in developmental disorders. Nature. 2017;542: 433–438. pmid:28135719
  8. 8. Kosmicki JA, Samocha KE, Howrigan DP, Sanders SJ, Slowikowski K, Lek M, et al. Refining the role of de novo protein-truncating variants in neurodevelopmental disorders by using population reference samples. Nat Genet. 2017;49: 504–510. pmid:28191890
  9. 9. Boyle EA, Li YI, Pritchard JK. An Expanded View of Complex Traits: From Polygenic to Omnigenic. Cell. 2017;169: 1177–1186. pmid:28622505
  10. 10. Jensen M, Girirajan S. Mapping a shared genetic basis for neurodevelopmental disorders. Genome Med. 2017;9: 109. pmid:29241461
  11. 11. Girirajan S, Eichler EE. Phenotypic variability and genetic susceptibility to genomic disorders. Hum Mol Genet. 2010;19: R176–87. pmid:20807775
  12. 12. Girirajan S, Rosenfeld JA, Coe BP, Parikh S, Friedman N, Goldstein A, et al. Phenotypic Heterogeneity of Genomic Disorders and Rare Copy-Number Variants. N Engl J Med. 2012;367: 1321–1331. pmid:22970919
  13. 13. Slager RE, Newton TL, Vlangos CN, Finucane B, Elsea SH. Mutations in RAI1 associated with Smith–Magenis syndrome. Nat Genet. 2003;33: 466–468. pmid:12652298
  14. 14. Bi W, Saifi GM, Shaw CJ, Walz K, Fonseca P, Wilson M, et al. Mutations of RAI1, a PHD-containing protein, in nondeletion patients with Smith-Magenis syndrome. Hum Genet. 2004;115: 515–524. pmid:15565467
  15. 15. Girirajan S, Vlangos CN, Szomju BB, Edelman E, Trevors CD, Dupuis L, et al. Genotype-phenotype correlation in Smith-Magenis syndrome: Evidence that multiple genes in 17p11.2 contribute to the clinical spectrum. Genet Med. 2006;8: 417–427. pmid:16845274
  16. 16. Bi W, Ohyama T, Nakamura H, Yan J, Visvanathan J, Justice MJ, et al. Inactivation of Rai1 in mice recapitulates phenotypes observed in chromosome engineered mouse models for Smith-Magenis syndrome. Hum Mol Genet. 2005;14: 983–995. pmid:15746153
  17. 17. Walz K, Caratini-Rivera S, Bi W, Fonseca P, Mansouri DL, Lynch J, et al. Modeling del(17)(p11.2p11.2) and dup(17)(p11.2p11.2) contiguous gene syndromes by chromosome engineering in mice: phenotypic consequences of gene dosage imbalance. Mol Cell Biol. 2003;23: 3646–3655. pmid:12724422
  18. 18. Adam MP, Ardinger HH, Pagon RA, Wallace SE, Bean LJ, Stephens K, et al. GeneReviews®. GeneReviews®. University of Washington, Seattle; 2018.
  19. 19. Weiss LA, Shen Y, Korn JM, Arking DE, Miller DT, Fossdal R, et al. Association between Microdeletion and Microduplication at 16p11.2 and Autism. N Engl J Med. 2008;358: 667–675. pmid:18184952
  20. 20. Willatt L, Cox J, Barber J, Cabanas ED, Collins A, Donnai D, et al. 3q29 microdeletion syndrome: clinical and molecular characterization of a new syndrome. Am J Hum Genet. 2005;77: 154–160. pmid:15918153
  21. 21. Bachmann-Gagescu R, Mefford HC, Cowan C, Glew GM, Hing A V, Wallace S, et al. Recurrent 200-kb deletions of 16p11.2 that include the SH2B1 gene are associated with developmental delay and obesity. Genet Med. 2010;12: 641–647. pmid:20808231
  22. 22. Brunetti-Pierri N, Berg JS, Scaglia F, Belmont J, Bacino CA, Sahoo T, et al. Recurrent reciprocal 1q21.1 deletions and duplications associated with microcephaly or macrocephaly and developmental and behavioral abnormalities. Nat Genet. 2008;40: 1466–1471. pmid:19029900
  23. 23. Cooper GM, Coe BP, Girirajan S, Rosenfeld JA, Vu TH, Baker C, et al. A copy number variation morbidity map of developmental delay. Nat Genet. 2011;43: 838–846. pmid:21841781
  24. 24. Girirajan S, Rosenfeld JA, Cooper GM, Antonacci F, Siswara P, Itsara A, et al. A recurrent 16p12.1 microdeletion supports a two-hit model for severe developmental delay. Nat Genet. 2010;42: 203–209. pmid:20154674
  25. 25. Rosenfeld JA, Coe BP, Eichler EE, Cuckle H, Shaffer LG. Estimates of penetrance for recurrent pathogenic copy-number variations. Genet Med. 2013;15: 478–481. pmid:23258348
  26. 26. Kirov G, Rees E, Walters JTR, Escott-Price V, Georgieva L, Richards AL, et al. The Penetrance of Copy Number Variations for Schizophrenia and Developmental Delay. Biol Psychiatry. 2014;75: 378–385. pmid:23992924
  27. 27. Stefansson H, Meyer-Lindenberg A, Steinberg S, Magnusdottir B, Morgen K, Arnarsdottir S, et al. CNVs conferring risk of autism or schizophrenia affect cognition in controls. Nature. 2014;505: 361–366. pmid:24352232
  28. 28. Kumar RA, Karamohamed S, Sudi J, Conrad DF, Brune C, Badner JA, et al. Recurrent 16p11.2 microdeletions in autism. Hum Mol Genet. 2008;17: 628–638. pmid:18156158
  29. 29. Shinawi M, Liu P, Kang SHL, Shen J, Belmont JW, Scott DA, et al. Recurrent reciprocal 16p11.2 rearrangements associated with global developmental delay, behavioural problems, dysmorphism, epilepsy, and abnormal head size. J Med Genet. 2010;47: 332–341. pmid:19914906
  30. 30. Zufferey F, Sherr EH, Beckmann ND, Hanson E, Maillard AM, Hippolyte L, et al. A 600 kb deletion syndrome at 16p11.2 leads to energy imbalance and neuropsychiatric disorders. J Med Genet. 2012;49: 660–668. pmid:23054248
  31. 31. Hanson E, Bernier R, Porche K, Jackson FI, Goin-Kochel RP, Snyder LG, et al. The Cognitive and Behavioral Phenotype of the 16p11.2 Deletion in a Clinically Ascertained Population. Biol Psychiatry. 2015;77: 785–793. pmid:25064419
  32. 32. Moreno-De-Luca A, Evans DW, Boomer KB, Hanson E, Bernier R, Goin-Kochel RP, et al. The role of parental cognitive, behavioral, and motor profiles in clinical variability in individuals with chromosome 16p11.2 deletions. JAMA Psychiatry. 2015;72: 119–126. pmid:25493922
  33. 33. Bochukova EG, Huang N, Keogh J, Henning E, Purmann C, Blaszczyk K, et al. Large, rare chromosomal deletions associated with severe early-onset obesity. Nature. 2010;463: 666–670. pmid:19966786
  34. 34. Walters RG, Jacquemont S, Valsesia A, de Smith AJ, Martinet D, Andersson J, et al. A new highly penetrant form of obesity due to deletions on chromosome 16p11.2. Nature. 2010;463: 671–675. pmid:20130649
  35. 35. Chaste P, Sanders SJ, Mohan KN, Klei L, Song Y, Murtha MT, et al. Modest impact on risk for autism spectrum disorder of rare copy number variants at 15q11.2, Specifically Breakpoints 1 to 2. Autism Res. 2014;7: 355–362. pmid:24821083
  36. 36. Pizzo L, Jensen M, Polyak A, Rosenfeld JA, Mannik K, Krishnan A, et al. Rare variants in the genetic background modulate cognitive and developmental phenotypes in individuals carrying disease-associated variants. Genet Med. 2018; 1–10. pmid:30190612
  37. 37. Golzio C, Katsanis N. Genetic architecture of reciprocal CNVs. Curr Opin Genet Dev. 2013;23: 240–248. pmid:23747035
  38. 38. Blaker-Lee A, Gupta S, McCammon JM, De Rienzo G, Sive H. Zebrafish homologs of genes within 16p11.2, a genomic region associated with brain disorders, are active during brain development, and include two deletion dosage sensor genes. Dis Model Mech. 2012;5: 834–851. pmid:22566537
  39. 39. Lindsay EA, Vitelli F, Su H, Morishima M, Huynh T, Pramparo T, et al. Tbx1 haploinsufficiency in the DiGeorge syndrome region causes aortic arch defects in mice. Nature. 2001;410: 97–101. pmid:11242049
  40. 40. Maynard TM, Haskell GT, Peters AZ, Sikich L, Lieberman JA, LaMantia A-S. A comprehensive analysis of 22q11 gene expression in the developing and adult brain. Proc Natl Acad Sci. 2003;100: 14433–14438. pmid:14614146
  41. 41. Prasad SE, Howley S, Murphy KC. Candidate genes and the behavioral phenotype in 22q11.2 deletion syndrome. Dev Disabil Res Rev. 2008;14: 26–34. pmid:18636634
  42. 42. Gillentine MA, Yin J, Bajic A, Zhang P, Cummock S, Kim JJ, et al. Functional Consequences of CHRNA7 Copy-Number Alterations in Induced Pluripotent Stem Cells and Neural Progenitor Cells. Am J Hum Genet. 2017;101: 874–887. pmid:29129316
  43. 43. Gillentine MA, Schaaf CP. The human clinical phenotypes of altered CHRNA7 copy number. Biochem Pharmacol. 2015;97: 352–362. pmid:26095975
  44. 44. Uddin M, Unda BK, Kwan V, Holzapfel NT, White SH, Chalil L, et al. OTUD7A Regulates Neurodevelopmental Phenotypes in the 15q13.3 Microdeletion Syndrome. Am J Hum Genet. 2018;102: 278–295. pmid:29395074
  45. 45. Yin J, Chen W, Chao ES, Soriano S, Wang L, Wang W, et al. Otud7a Knockout Mice Recapitulate Many Neurological Features of 15q13.3 Microdeletion Syndrome. Am J Hum Genet. 2018;102: 296–308. pmid:29395075
  46. 46. Escamilla CO, Filonova I, Walker AK, Xuan ZX, Holehonnur R, Espinosa F, et al. Kctd13 deletion reduces synaptic transmission via increased RhoA. Nature. 2017;551: 227–231. pmid:29088697
  47. 47. Iyer J, Singh MD, Jensen M, Patel P, Pizzo L, Huber E, et al. Pervasive genetic interactions modulate neurodevelopmental defects of the autism-associated 16p11.2 deletion in Drosophila melanogaster. Nat Commun. 2018;9: 2548. pmid:29959322
  48. 48. McCammon JM, Blaker-Lee A, Chen X, Sive H. The 16p11.2 homologs fam57ba and doc2a generate certain brain and body phenotypes. Hum Mol Genet. 2017;26: 3699–3712. pmid:28934389
  49. 49. Ip JPK, Nagakura I, Petravicz J, Li K, Wiemer EAC, Sur M. Major vault protein, a candidate gene in 16p11.2 microdeletion syndrome, is required for the homeostatic regulation of visual cortical plasticity. J Neurosci. 2018;38: 2034–17. pmid:29540554
  50. 50. Yadav S, Oses-Prieto JA, Peters CJ, Zhou J, Pleasure SJ, Burlingame AL, et al. TAOK2 Kinase Mediates PSD95 Stability and Dendritic Spine Maturation through Septin7 Phosphorylation. Neuron. 2017;93: 379–393. pmid:28065648
  51. 51. Richter M, Murtaza N, Scharrenberg R, White SH, Johanns O, Walker S, et al. Altered TAOK2 activity causes autism-related neurodevelopmental and cognitive abnormalities through RhoA signaling. Mol Psychiatry. 2018; 1–22. pmid:29467497
  52. 52. Park SM, Littleton JT, Park HR, Lee JH. Drosophila Homolog of Human KIF22 at the Autism-Linked 16p11. 2 Loci Influences Synaptic Connectivity at Larval Neuromuscular Junctions. Exp Neurobiol. 2016;25: 33–39. pmid:26924931
  53. 53. Loviglio MN, Arbogast T, Jønch AE, Collins SC, Popadin K, Bonnet CS, et al. The Immune Signaling Adaptor LAT Contributes to the Neuroanatomical Phenotype of 16p11.2 BP2-BP3 CNVs. Am J Hum Genet. 2017;101: 564–577. pmid:28965845
  54. 54. Wu N, Ming X, Xiao J, Wu Z, Chen X, Shinawi M, et al. TBX6 Null Variants and a Common Hypomorphic Allele in Congenital Scoliosis. N Engl J Med. 2015;372: 341–350. pmid:25564734
  55. 55. Gong W, Gottlieb S, Collins J, Blescia A, Dietz H, Goldmuntz E, et al. Mutation analysis of TBX1 in non-deleted patients with features of DGS/VCFS or isolated cardiovascular defects. J Med Genet. 2001;38: E45. pmid:11748311
  56. 56. Ponnam SPG, Ramesha K, Tejwani S, Ramamurthy B, Kannabiran C. Mutation of the gap junction protein alpha 8 (GJA8) gene causes autosomal recessive cataract. J Med Genet. 2007;44: e85–e85. pmid:17601931
  57. 57. Soemedi R, Topf A, Wilson IJ, Darlay R, Rahman T, Glen E, et al. Phenotype-specific effect of chromosome 1q21.1 rearrangements and GJA5 duplications in 2436 congenital heart disease patients and 6760 controls. Hum Mol Genet. 2012;21: 1513–1520. pmid:22199024
  58. 58. Pannu H, Tran-Fadulu V, Papke CL, Scherer S, Liu Y, Presley C, et al. MYH11 mutations result in a distinct vascular pathology driven by insulin-like growth factor 1 and angiotensin II. Hum Mol Genet. 2007;16: 2453–2462. pmid:17666408
  59. 59. Kuang SQ, Guo DC, Prakash SK, McDonald MLN, Johnson RJ, Wang M, et al. Recurrent chromosome 16p13.1 duplications are a risk factor for aortic dissections. PLoS Genet. 2011;7: e1002118. pmid:21698135
  60. 60. Girirajan S, Campbell CD, Eichler EE. Human Copy Number Variation and Complex Genetic Disease. Annu Rev Genet. 2011;45: 203–226. pmid:21854229
  61. 61. Huang N, Lee I, Marcotte EM, Hurles ME. Characterising and predicting haploinsufficiency in the human genome. PLoS Genet. 2010;6:e1001154. pmid:20976243
  62. 62. Steinberg J, Honti F, Meader S, Webber C. Haploinsufficiency predictions without study bias. Nucleic Acids Res. 2015;43: e101. pmid:26001969
  63. 63. Khurana E, Fu Y, Chen J, Gerstein M. Interpretation of Genomic Variants Using a Unified Biological Network Approach. PLoS Comput Biol. 2013;9: e1002886. pmid:23505346
  64. 64. Han X, Chen S, Flynn E, Wu S, Wintner D, Shen Y. Distinct epigenomic patterns are associated with haploinsufficiency and predict risk genes of developmental disorders. Nat Commun. 2018;9: 2138. pmid:29849042
  65. 65. Lek M, Karczewski KJ, Minikel E V., Samocha KE, Banks E, Fennell T, et al. Analysis of protein-coding genetic variation in 60,706 humans. Nature. 2016;536: 285–291. pmid:27535533
  66. 66. Petrovski S, Wang Q, Heinzen EL, Allen AS, Goldstein DB. Genic Intolerance to Functional Variation and the Interpretation of Personal Genomes. PLoS Genet. 2013;9: e1003709. pmid:23990802
  67. 67. Havrilla JM, Pedersen BS, Layer RM, Quinlan AR. A map of constrained coding regions in the human genome; 2017. Preprint. Available in: bioRxiv: 220814.
  68. 68. Krishnan A, Zhang R, Yao V, Theesfeld CL, Wong AK, Tadych A, et al. Genome-wide prediction and functional characterization of the genetic basis of autism spectrum disorder. Nat Neurosci. 2016;19: 1454–1462. pmid:27479844
  69. 69. Pucilowska J, Vithayathil J, Tavares EJ, Kelly C, Karlo JC, Landreth GE. The 16p11.2 Deletion Mouse Model of Autism Exhibits Altered Cortical Progenitor Proliferation and Brain Cytoarchitecture Linked to the ERK MAPK Pathway. J Neurosci. 2015;35: 3190–3200. pmid:25698753
  70. 70. Arbogast T, Ouagazzal A-M, Chevalier C, Kopanitsa M, Afinowi N, Migliavacca E, et al. Reciprocal Effects on Neurocognitive and Metabolic Phenotypes in Mouse Models of 16p11.2 Deletion and Duplication Syndromes. PLoS Genet. 2016;12: e1005709. pmid:26872257
  71. 71. Portmann T, Yang M, Mao R, Panagiotakos G, Ellegood J, Dolen G, et al. Behavioral abnormalities and circuit defects in the basal ganglia of a mouse model of 16p11.2 deletion syndrome. Cell Rep. 2014;7: 1077–1092. pmid:24794428
  72. 72. Horev G, Ellegood J, Lerch JP, Son Y-EE, Muthuswamy L, Vogel H, et al. Dosage-dependent phenotypes in models of 16p11.2 lesions found in autism. Proc Natl Acad Sci. 2011;108: 17076–17081. pmid:21969575
  73. 73. Blizinsky KD, Diaz-Castro B, Forrest MP, Schürmann B, Bach AP, Martin-de-Saavedra MD, et al. Reversal of dendritic phenotypes in 16p11.2 microduplication mouse model neurons by pharmacological targeting of a network hub. Proc Natl Acad Sci. 2016;113: 8520–8525. pmid:27402753
  74. 74. Golzio C, Willer J, Talkowski ME, Oh EC, Taniguchi Y, Jacquemont S, et al. KCTD13 is a major driver of mirrored neuroanatomical phenotypes of the 16p11.2 copy number variant. Nature. 2012;485: 363–367. pmid:22596160
  75. 75. Mazzucchelli C, Vantaggiato C, Ciamei A, Fasano S, Pakhotin P, Krezel W, et al. Knockout of ERK1 MAP kinase enhances synaptic plasticity in the striatum and facilitates striatal-mediated learning and memory. Neuron. 2002;34: 807–20. pmid:12062026
  76. 76. Voisin L, Saba-El-Leil MK, Julien C, Fremin C, Meloche S. Genetic Demonstration of a Redundant Role of Extracellular Signal-Regulated Kinase 1 (ERK1) and ERK2 Mitogen-Activated Protein Kinases in Promoting Fibroblast Proliferation. Mol Cell Biol. 2010;30: 2918–2932. pmid:20368360
  77. 77. Haller M, Au J, O’Neill M, Lamb DJ. 16p11.2 transcription factor MAZ is a dosage-sensitive regulator of genitourinary development. Proc Natl Acad Sci. 2018;115: 201716092. pmid:29432158
  78. 78. Ohsugi M, Adachi K, Horai R, Kakuta S, Sudo K, Kotaki H, et al. Kid-Mediated Chromosome Compaction Ensures Proper Nuclear Envelope Formation. Cell. 2008;132: 771–782. pmid:18329364
  79. 79. Michetti C, Castroflorio E, Marchionni I, Forte N, Sterlini B, Binda F, et al. The PRRT2 knockout mouse recapitulates the neurological diseases associated with PRRT2 mutations. Neurobiol Dis. 2017;99: 66–83. pmid:28007585
  80. 80. Kumar A, Lualdi M, Loncarek J, Cho YW, Lee JE, Ge K, et al. Loss of function of mouse Pax-Interacting Protein 1-associated glutamate rich protein 1a (Pagr1a) leads to reduced Bmp2 expression and defects in chorion and amnion development. Dev Dyn. 2014;243: 937–947. pmid:24633704
  81. 81. Lee C-M, He CH, Nour AM, Zhou Y, Ma B, Park JW, et al. IL-13Rα2 uses TMEM219 in chitinase 3-like-1-induced signalling and effector responses. Nat Commun. 2016;7: 12752. pmid:27629921
  82. 82. Sakaguchi G, Manabe T, Kobayashi K, Orita S, Sasaki T, Naito A, et al. Doc2α is an activity-dependent modulator of excitatory synaptic transmission. Eur J Neurosci. 1999;11: 4262–4268. pmid:10594652
  83. 83. Shui J-W, Hu MC-T, Tan T-H. Conditional Knockout Mice Reveal an Essential Role of Protein Phosphatase 4 in Thymocyte Development and Pre-T-Cell Receptor Signaling. Mol Cell Biol. 2007;27: 79–91. pmid:17060460
  84. 84. Toyo-oka K, Mori D, Yano Y, Shiota M, Iwao H, Goto H, et al. Protein phosphatase 4 catalytic subunit regulates Cdk1 activity and microtubule organization via NDEL1 dephosphorylation. J Cell Biol. 2008;180: 1133–1147. pmid:18347064
  85. 85. Chapman DL, Papaioannou VE. Three neural tubes in mouse embryos with mutations in the T-box gene Tbx6. Nature. 1998;391: 695–697. pmid:9490412
  86. 86. Takemoto T, Uchikawa M, Yoshida M, Bell DM, Lovell-Badge R, Papaioannou VE, et al. Tbx6-dependent Sox2 regulation determines neural or mesodermal fate in axial stem cells. Nature. 2011;470: 394–398. pmid:21331042
  87. 87. Pagès G, Guérin S, Grall D, Bonino F, Smith A, Anjuere F, et al. Defective thymocyte maturation in p44 MAP kinase (Erk 1) knockout mice. Science. 1999;286: 1374–1377. pmid:10558995
  88. 88. Forsingdal A, Fejgin K, Nielsen V, Werge T, Nielsen J. 15q13.3 homozygous knockout mouse model display epilepsy-, autism- and schizophrenia-related phenotypes. Transl Psychiatry. 2016;6: e860. pmid:27459725
  89. 89. Yin J, Chen W, Yang H, Xue M, Schaaf CP. Chrna7 deficient mice manifest no consistent neuropsychiatric and behavioral phenotypes. Sci Rep. 2017;7: 39941. pmid:28045139
  90. 90. Smith CL, Blake JA, Kadin JA, Richardson JE, Bult CJ, Mouse Genome Database Group. Mouse Genome Database (MGD)-2018: knowledgebase for the laboratory mouse. Nucleic Acids Res. 2018;46: D836–D842. pmid:29092072
  91. 91. Howe DG, Bradford YM, Conlin T, Eagle AE, Fashena D, Frazer K, et al. ZFIN, the Zebrafish Model Organism Database: increased support for mutants and transgenics. Nucleic Acids Res. 2012;41: D854–D860. pmid:23074187
  92. 92. Gramates LS, Marygold SJ, Dos Santos G, Urbano JM, Antonazzo G, Matthews BB, et al. FlyBase at 25: Looking to the future. Nucleic Acids Res. 2017;45: D663–D671. pmid:27799470
  93. 93. Langfelder P, Horvath S. WGCNA: An R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9: 559. pmid:19114008
  94. 94. Miller JA, Ding S-L, Sunkin SM, Smith KA, Ng L, Szafer A, et al. Transcriptional landscape of the prenatal human brain. Nature. 2014;508: 199–206. pmid:24695229
  95. 95. Chatr-aryamontri A, Oughtred R, Boucher L, Rust J, Chang C, Kolas NK, et al. The BioGRID interaction database: 2017 update. Nucleic Acids Res. 2017;45: D369–D379. pmid:27980099
  96. 96. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks. Genome Res. 2003;13: 2498–2504. pmid:14597658
  97. 97. Mi H, Muruganujan A, Casagrande JT, Thomas PD. Large-scale gene function analysis with the PANTHER classification system. Nat Protoc. 2013;8: 1551–1566. pmid:23868073
  98. 98. Hu T, Yamagishi H, Maeda J, McAnally J, Yamagishi C, Srivastava D. Tbx1 regulates fibroblast growth factors in the anterior heart field through a reinforcing autoregulatory loop involving forkhead transcription factors. Development. 2004;131: 5491–502. pmid:15469978
  99. 99. Vitelli F, Zhang Z, Huynh T, Sobotka A, Mupo A, Baldini A. Fgf8 expression in the Tbx1 domain causes skeletal abnormalities and modifies the aortic arch but not the outflow tract phenotype of Tbx1 mutants. Dev Biol. 2006;295: 559–70. pmid:16696966
  100. 100. Satoh Y, Endo S, Ikeda T, Yamada K, Ito M, Kuroki M, et al. Extracellular signal-regulated kinase 2 (ERK2) knockdown mice show deficits in long-term memory; ERK2 has a specific function in learning and memory. J Neurosci. 2007;27: 10765–76. pmid:17913910
  101. 101. Landrum MJ, Lee JM, Benson M, Brown GR, Chao C, Chitipiralla S, et al. ClinVar: improving access to variant interpretations and supporting evidence. Nucleic Acids Res. 2018;46: D1062–D1067. pmid:29165669
  102. 102. Turner TN, Yi Q, Krumm N, Huddleston J, Hoekzema K, F Stessman HA, et al. denovo-db: a compendium of human de novo variants. Nucleic Acids Res. 2017;45: D804–D811. pmid:27907889
  103. 103. Gonzalez-Mantilla AJ, Moreno-De-Luca A, Ledbetter DH, Martin CL. A cross-disorder method to identify novel candidate genes for developmental brain disorders. JAMA Psychiatry. 2016;73: 275–283. pmid:26817790
  104. 104. Ruderfer DM, Hamamsy T, Lek M, Karczewski KJ, Kavanagh D, Samocha KE, et al. Patterns of genic intolerance of rare copy number variation in 59,898 human exomes. Nat Genet. 2016;48: 1107–1111. pmid:27533299
  105. 105. Andrews T, Honti F, Pfundt R, de Leeuw N, Hehir-Kwa J, Vulto-van Silfhout A, et al. The clustering of functionally related genes contributes to CNV-mediated disease. Genome Res. 2015;25: 802–13. pmid:25887030
  106. 106. Noh HJ, Ponting CP, Boulding HC, Meader S, Betancur C, Buxbaum JD, et al. Network Topologies and Convergent Aetiologies Arising from Deletions and Duplications Observed in Individuals with Autism. PLoS Genet. 2013;9: e1003523. pmid:23754953
  107. 107. Grice SJ, Liu JL, Webber C. Synergistic Interactions between Drosophila Orthologues of Genes Spanned by De Novo Human CNVs Support Multiple-Hit Models of Autism. PLoS Genet. 2015;11: e1004998. pmid:25816101
  108. 108. Mitchell KJ. The genetics of neurodevelopmental disease. Curr Opin Neurobiol. 2011;21: 197–203. pmid:20832285
  109. 109. Hoppman-Chaney N, Wain K, Seger P, Superneau D, Hodge J. Identification of single gene deletions at 15q13.3: further evidence that CHRNA7 causes the 15q13.3 microdeletion syndrome phenotype. Clin Genet. 2013;83: 345–351. pmid:22775350
  110. 110. Gokhale A, Freeman AA, Hartwig C, Bassell JL, Zlatic SA, Sapp C, et al. Systems Analysis of the 22q11.2 Microdeletion Syndrome Converges on a Mitochondrial Interactome Necessary for Synapse Function and Behavior; 2018. Preprint. Available in: bioRxiv: 315143.