ITAN, the Italian Autism Network. List of ITAN author names are Carmela Bravaccio, Paolo Curatolo, Lucio Da Ros, Bernardo Della Bernardina, Maurizio Elia, Serenalli Grittani, Lucia Margari, Gabriele Masi, Massimo Molteni, Pierfranco Pignatti, Paola Prandini, Alessandra Tiberti, Elisabetta Trabetti, Leonardo Zoccante, Alessandro Zuddas
No association between a common single nucleotide polymorphism, rs4141463, in the MACROD2 gene and autism spectrum disorder†
Article first published online: 8 JUN 2011
Copyright © 2011 Wiley-Liss, Inc.
American Journal of Medical Genetics Part B: Neuropsychiatric Genetics
Volume 156, Issue 6, pages 633–639, September 2011
How to Cite
Curran, S., Bolton, P., Rozsnyai, K., Chiocchetti, A., Klauck, S. M., Duketis, E., Poustka, F., Schlitt, S., Freitag, C. M., Lee, I., Muglia, P., on behalf of the ITAN, Poot, M., Staal, W., de Jonge, M. V., Ophoff, R. A., Lewis, C., Skuse, D., Mandy, W., Vassos, E., Fossdal, R., Magnusson, P., Hreidarsson, S., Saemundsen, E., Stefansson, H., Stefansson, K. and Collier, D. (2011), No association between a common single nucleotide polymorphism, rs4141463, in the MACROD2 gene and autism spectrum disorder. Am. J. Med. Genet., 156: 633–639. doi: 10.1002/ajmg.b.31201
How to Cite this Article: Curran S, Bolton P, Rozsnyai K, Chiocchetti A, Klauck SM, Duketis E, Poustka F, Schlitt S, Freitag CM, Lee I, Muglia P (on behalf of the ITAN), Poot M, Staal W, de Jonge MV, Ophoff R A, Lewis C, Skuse D, Mandy W, Vassos E, Fossdal R, Magnusson P, Hreidarsson S, Saemundsen E, Stefansson H, Stefansson K, Collier D. 2011. No Association Between a Common Single Nucleotide Polymorphism, rs4141463, in the MACROD2 Gene and Autism Spectrum Disorder. Am J Med Genet Part B 156:633–639.
- Issue published online: 10 AUG 2011
- Article first published online: 8 JUN 2011
- Manuscript Accepted: 25 APR 2011
- Manuscript Received: 8 DEC 2010
- UK Medical Research Council. Grant Number: G0500079
- Deutsche Forschungsgemeinschaft
- EU Grant PsychCNV. Grant Number: HEALTH-2007-2.2.1-10-223423
- NIH. Grant Number: MH071425
- The Netherlands Foundation for Brain Research (Hersenstichting). Grant Numbers: 2008(1).34, F2008(1)
- autism spectrum;
- genetic association;
- common genetic risk variants;
The Autism Genome Project (AGP) Consortium recently reported genome-wide significant association between autism and an intronic single nucleotide polymorphism marker, rs4141463, within the MACROD2 gene. In the present study we attempted to replicate this finding using an independent case–control design of 1,170 cases with autism spectrum disorder (ASD) (874 of which fulfilled narrow criteria for Autism (A)) from five centers within Europe (UK, Germany, the Netherlands, Italy, and Iceland), and 35,307 controls. The combined sample size gave us a non-centrality parameter (NCP) of 11.9, with 93% power to detect allelic association of rs4141463 at an alpha of 0.05 with odds ratio of 0.84 (the best odds ratio estimate of the AGP Consortium data), and for the narrow diagnosis of autism, an NCP of 8.9 and power of 85%. Our case–control data were analyzed for association, stratified by each center, and the summary statistics were combined using the meta-analysis program, GWAMA. This resulted in an odds ratio (OR) of 1.03 (95% CI 0.944–1.133), with a P-value of 0.5 for ASD and OR of 0.99 (95% CI 0.88–1.11) with P-value = 0.85 for the Autism (A) sub-group. Therefore, this study does not provide support for the reported association between rs4141463 and autism. © 2011 Wiley-Liss, Inc.
There is compelling evidence from twin, family, and population-based studies that genetic factors are the most significant contributors to the etiology of autism spectrum disorders (ASD) and related traits [Freitag, 2007], with a heritability, estimated to be about 70–90% [Skuse, 2007] among the highest observed for any common neuropsychiatric disorder. The genetic architecture of ASD is, however, likely to be complex and its exact structure remains elusive [El-Fishawy and State, 2010]. Previously, the most widely accepted hypothesis for the genetic basis of common disorders was the “common variant” hypothesis, in which many common, low risk (odds ratio—OR < 1.5) alleles act together to increase disease risk [Risch and Merikangas, 1996]. However, it has recently become evident that rare, moderate risk variants such as de novo copy number variants are an important contributor to complex disease risk, especially for autism and schizophrenia [Miller et al., 2010]. These comprise microscopically observable high risk cytogenetic abnormalities (e.g., maternal duplication of 15q11–q13) (∼1–3% of ASD subjects), and single-gene syndromes (e.g., tuberous sclerosis TSC1, TSC2), which have ASD phenotypes, and together account for about 10% of ASD cases [Abrahams and Geschwind, 2008]. In addition a number of moderate risk copy number variants (CNVs) have been associated with autism found thus far in about 10% of cases [Sebat et al., 2007; Szatmari et al., 2007; Christian et al., 2008; Marshall et al., 2008; Glessner et al., 2009]. In a few patients point mutations have also been identified by re-sequencing, such as those found in SHANK3 [Gauthier et al., 2009]. However, in the majority of ASD cases (>80%), there is no identifiable genetic cause or genomic risk factor, such as an autism-associated CNV or chromosomal abnormality, and overall, the genetic heterogeneity is expected to be substantial, thus hampering genetic studies [Coon et al., 2010].
While empirical observations currently favor a major role for rare, moderate risk variants in autism, common, low risk genetic variants are also expected to be involved, as is the case for all other complex genetic disorders studied so far which have undergone GWAS (genome-wide association study) analysis [Cichon et al., 2009]. Common variants might account for some of the “dark matter” or “missing heritability” [Manolio et al., 2009] of undiscovered susceptibility alleles, and therefore it is important to carefully assess GWAS association data, as it can provide important information on genetic architecture. Evidence for one genome-wide significant GWAS association in autism with the single nucleotide polymorphism rs4141463, located in an intron of the MACROD2 gene, comes from a family-based GWAS study performed by the AGP Consortium, in which a total of 1,385 ASD probands from 1,369 families were analyzed [Anney et al., 2010]. The strongest signal for association, reaching genome-wide significance at P = 2.1 × 10−8 (OR 0.56; 95% CI 0.47–0.67) was for rs4141463 (C/T) in a sub-set of 720 ASD probands (from 718 families) with a strictly defined ASD diagnosis and European ancestry, in which the common European allele rs4141463C (dbSNP build 131, RefSNP alleles: C/T; ancestral allele: T; ss126770843) was over transmitted in cases. In a replication sample in the same study, a somewhat higher but non-significant OR was observed (P = 0.13; OR 0.84; 95% CI 0.67–1.04). Here, we aimed to replicate the reported association of rs4141463 in a large, independent sample of individuals with ASD with European ancestry.
MATERIALS AND METHODS
A sample of 1,170 ASD subjects of European origin was collected from centers in South-East England (MGAS, UCL, and TEDS cohorts), Germany (Frankfurt/Heidelberg), the Netherlands (UMC Utrecht), Italy, and Iceland and genotyped. This study was approved by the National Bioethics Committees or the Local Research Ethical Committees, and was compliant with Data Protection Commissions or laws in each country. Written consent was obtained from all individuals, their and assent or consent from parents when appropriate. All cases met criteria for autism, Asperger's syndrome, or pervasive developmental disorder, not otherwise specified (PDD-NOS) according to the DSM-IV classification system using a standard research assessment protocol that included the autism diagnostic interview-revised (ADI-R) [Lord et al., 1994] and autism diagnostic observation schedule (ADOS) [Lord et al., 1989] and was diagnostically confirmed by clinician consensus [Mazefsky and Oswald, 2006]. This methodology is very similar to the study by Anney et al. . Eight hundred seventy-four out of 1,032 (85%) subjects were classified as fulfilling narrow autism criteria. Excluded were ASD subjects with identified metabolic, genetic syndromes, or progressive neurological disorders and profound intellectual disability (IQ/DQ < 20). Each center had ethnically matched controls. Genotyping was performed using TaqMan SNP Genotyping Assays. Quality control measures included placing 3 blank negative control samples on each 384-well plate and a minimum of 4 duplicated samples for testing inter-plate reproducibility. Samples from UK and NL were genotyped in the same lab and cross-validation of UK and German genotyping calls were made by genotyping 12 UK samples at the German site with perfect matching of results.
Power was calculated using the genetic power calculator (GPC) [Purcell et al., 2003], with parameters of the protective allele frequency (A), 0.42; prevalence, 0.01 and genotypic relative risk (GRR) for Aa, 0.85 and AA, 0.72 which is equivalent to the best previous estimator of the OR, derived from the independent sample evaluated by the AGP, namely the AGRE sample [Anney et al., 2010]. Our combined sample size of 1,170, according to GPC, gave us a non-centrality parameter (NCP) of 11.9, with 93% power to detect a common genetic variant at a significance level (alpha) of 0.05 and 81% at 0.01. With the 874 narrowly defined cases of autism, power fell to 85% at alpha = 0.05 and 66% at alpha = 0.01. Power was 99% for detection of a GRR of 0.65, as in the AGP discovery sample, although as the authors explain, the OR of 0.65 is probably exaggerated due to scanning the entire genome for significant findings.
The results of case–control association tests for each center are presented in Table I, where data for ASD and strict autism are both presented, and Figure 1. Genotypes were checked for HWE using http://www.oege.org/software/hwe-mr-calc.shtml (UK controls, χ2 1.97, P = 0.3734; Germany controls, χ2 = 0.95, P = 0.6219; the Netherlands controls, χ2 = 1.47, P = 0.4795; Italy controls χ2 = 1.08, P = 0.5827). One data set (Iceland) had genome control data. As our data came from geographically distinct regions, we combined the summary statistics from each center, analyzed separately, using the meta-analysis software tool, GWAMA [genome-wide association meta-analysis; Magi and Morris, 2010] http://www.well.ox.ac.uk/GWAMA. Combining the association data for all ASD subjects in our study, the GWAMA output (Table II) gave an OR of 1.03 (95% CI 0.944–1.133) with P = 0.5. The test of heterogeneity between samples gave a Q statistic of 9.68 (P = 0.05) and I2 = 0.587, indicating that there was moderate heterogeneity between the samples, although there were only five different populations. On the one hand, given that there were only five centers, this level of heterogeneity might be noteworthy, although power to detect heterogeneity may be low in this number of groups. The German sample reached nominal significance for ASD (P = 0.029) and narrowly defined autism (P = 0.039). The meta-analysis results using data from the subjects meeting strict autism criteria were similar (Table III). Here the OR was 0.99 (95% CI 0.88–1.11) with P-value = 0.85. There was somewhat less heterogeneity between samples detected by GWAMA (the Q statistic was 7.05 (P = 0.133) and I2 = 0.43).
|Data set||N (ASD = 1,170; autism = 760; controls = 35,307)||CC||CT||TT||C||T||Maf (T)||Odds ratio for C allele 95% confidence intervals||Chi-sq||P-value|
|UK||ASD = 324||104||176||44||384||264||0.41||ASD = 1.19 (0.99–1.43)||3.42||0.064|
|Autism = 237||71||128||38||270||204||0.43||A = 1.08 (0.88–1.33)||0.57||0.45|
|Controls = 820||258||386||176||902||738||0.45|
|Germany||ASD = 236||71||113||52||255||217||0.46||ASD = 0.80 (0.66–0.98)||4.79||0.029|
|Autism = 225||69||106||50||244||206||0.46||A = 0.807 (0.65–0.98)||4.28||0.039|
|Controls = 1145||413||536||196||1362||928||0.41||ASD = 1.17 (0.93–1.48)||1.72||0.19|
|(Heidelberg = 174||58||88||28||204||144||0.41||A = 1.22 (0.94–1.58)||2.27||0.13|
|Bonn = 367||130||176||61||436||298||0.41|
|Munich = 604)||225||272||107||722||486||0.40|
|NL||ASD = 184||65||89||30||219||149||0.41|
|Autism = 149||53||73||22||179||117||0.40|
|(Poot ASD = 46||18||19||9||55||37||0.40|
|Poot autism = 44||17||19||8||53||35||0.40|
|Staal ASD = 138||47||70||21||164||112||0.41|
|Staal autism = 105)||36||54||14||126||82||0.40|
|Controls = 628||202||295||131||699||557||0.44|
|Italy||ASD = 98||35||37||26||107||89||0.45||ASD = 1.00(0.67–1.51)||0||1|
|Autism = 90||33||33||14||99||81||0.45||A = 1.02 (0.67–1.51)||0.01||0.92|
|Controls = 89||24||49||16||97||81||0.46|
|Iceland||ASD = 328||360||296||0.45||ASD = 1.04 (0.89–1.21)||0.25||0.62|
|Autism = 173||192||154||0.45||A = 1.06 (0.86–1.31)||0.31||0.58|
|Controls = 32625||35238||30012||0.46|
|Controls = 32625||35238||30012||0.46|
|rs number||Reference allele||Other allele||OR||OR_95L||OR_95U||P-value||Q statistic and P-value||I2|
|rs number||Reference allele||Other allele||OR||OR_95L||OR_95U||P-value||Q statistic and P-value||I2|
The purpose of this study was to attempt to replicate a reported genome-wide significant association between ASD and a common genetic variant in an intron of MACROD2, rs4141463, from a GWAS study [Anney et al., 2010]. However, we were unable to replicate or provide support for the original association, as our OR was close to 1, consistent with the null hypothesis. The potential reasons for the non-replication of genetic association findings are many, and include the four most obvious explanations, that (i) the original association was a chance false-positive finding, (ii) the association was caused by population stratification in the original study, (iii) there was a technical genotyping artifact, or (iv) the replication study was underpowered, that is, there was a lack of statistical power [Redden and Allison, 2003]. Even with genome-wide significant P-values of 2 × 10−8 or lower (as in the original study) there is the possibility of false positive association. The established 2 × 10−8 threshold roughly equals the P = 0.05 genome-wide. With such a level there is still 5% chance of finding false positives. To avoid this possibility the requirement that genome-wide significance should be reached, at least once, in a single study population may be appropriate. However, this will be difficult to achieve with current sample sizes. Population stratification seems unlikely in the analysis by Anney et al.  since both family-based analysis (which will avoid population stratification at the cost of power) and case–control analysis were used. There is no evidence of a genotyping artifact with rs4141463; as it is not an isolated SNP, with neighboring SNPs showing evidence of association and it is not in a repetitive DNA segment (UCSC browser, chr20:14,747,221–14,747,721, February 2009 (GRCh37/hg19) assembly).
However, rare CNVs are known at the MACROD2 locus, which is a deletion hotspot [Bradley et al., 2010], although these seem too rare to cause genotyping problems. One strand of supportive evidence for a role for the MACROD2 gene in autism is a rare deletion of the locus, seen in a case of Kabuki syndrome [Maas et al., 2007], which can feature autistic-like symptoms [Ho and Eaves, 1997; Akin Sari et al., 2008] and one individual with schizophrenia [Xu et al., 2009], although the deletion was not seen in 43 other patients with Kabuki syndrome [Kuniba et al., 2008]. Without clear evidence of disease association to date, MACROD2 deletions are thought to be probably non-pathogenic [Bradley et al., 2010].
In the manuscript by Anney et al. , rs4141463 and other SNPs reaching significance of P < 5 × 10−6 were selected for replication in an independent sample of 1,086 ASD probands from 595 families (∼50% with a strict autism diagnosis) from the Autism Genetics Resource Exchange (AGRE) database. The association with rs4141463 did not replicate in this sample (P = 0.13, OR 0.84 [CI 0.67–1.04]). Neither of the other recent GWAS studies [Arking et al., 2009; Ma et al., 2009; Wang et al., 2009; Weiss et al., 2009] found evidence for implicating MACROD2 or nearby variants in ASD, despite each study testing hundreds of cases in multiplex families or thousands of ASD subject cases and controls of European ancestry. Interestingly, a key resource for the Wang, Weiss, and Anney studies is the AGRE data resource, so these three papers do not describe three independent pieces of information sets of data. Thus, it is difficult to interpret the reasons why some of these samples found association and some not. The reasons these studies have obtained considerably differing results may be due to genotyping differences (differing genotyping platforms), different data cleaning procedures and statistical analyses, and to the use of partly different sub-samples of the AGRE sampler source.
Our sample size is large enough to detect a significant association (at P < 0.05) for a risk variant with a genotype relative risk of the same order as the best estimate from Anney et al. , 0.84, and has even greater power to detect the AGP (OR < 0.56) or AGP + AGRE (OR < 0.65) samples. However, there may still have been an overestimation of the genetic effect in that first study (“winners curse”), and if the true OR is closer to 1 (e.g., GRR 0.9) then the power in our sample will be less, at 48% for alpha = 0.05. There is also disagreement about what constitutes an adequate replication or refutation [Chanock et al., 2007], which can range from a null result, a directional effect, that is, the allelic association being in the same direction as the original observation albeit not significant, through to a GWAS-significance replication level of association at <5 × 10−8. In the present study, our OR of 1.03 is significantly higher than that of Anney et al.  and our 95% confidence intervals for a GRR of 0.94–1.13 do not cross the original GRR of 0.65. However, although our results do not support association, we cannot formally exclude a role for rs4141463 in risk of developing autism. It might, for example, be a proxy for a pathogenic variant that is rare or has varying linkage disequilibrium with rs4141463 according to geographic origin of the study population, making it difficult to consistently replicate.
There were differences in the research design of the cohorts studied; the AGP and AGRE samples are family trios, with control cases, and ours was a case–control study. The diagnostic criteria (ADI-R diagnosis of autism or ASD) and population types (Northern European) were, however, comparable in both studies. One cannot rule out the possibility that the original association, despite reaching genome-wide significance, is a false positive association. By looking closer to the study by Anney et al. , we observe that the original finding was not supported in their second sample (AGRE) (P = 0.13). In the overall sample (AGP + AGRE + SAGE) the association retained genome-wide significance despite the increase in the OR from 0.56 to 0.73 due to the increase in the overall sample size.
Several other potential common, low risk variants for autism have been reported to date; these are typical of complex disease risk alleles detected by GWAS in that they have ORs of <1.3 [Pawitan et al., 2009]. These GWAS results also appear to rule out moderate risk common variants (i.e., increasing relative risk by twofold or more), as is the case for most other common, complex diseases. Three such potential loci for ASD have been identified: rs4307059 on 5p14.1, between neuronal cadherin genes CDH9 and CDH10 [Wang et al., 2009], rs10513025 on 5p15.2, between the semaphorin (SEMA5A) and bitter taste receptor (TAS2R1) genes [Weiss et al., 2009] and MACROD2 [Anney et al., 2010]. In addition there are many other reported associations from candidate gene analysis, which have variable levels of evidence for association. Thus, at present there are no strongly associated common low risk alleles known, and further replication is required. However, at the current available GWAS sample size for autism, common, low risk variants have been identified for other complex disorders, such as obesity [Peterson et al., 2011].
Thus, a major role for common, low risk variants in autism remains unproven. Quantitative genetic studies point towards the involvement of many common variants, together influencing ASD or sub-domains of ASD [Ronald et al., 2010]. Since common variants are generally of small effect size [median OR < 1.25; Ralston, 2010] then if the heritability of autism were predominantly explained by these, hundreds of risk variants would need to exist [Pawitan et al., 2009], and many thousands, even hundreds of thousands, of cases would be required to detect them. The extent of the role of common variants will only be known once better-powered GWAS studies are performed, and the results replicated. In obesity, current GWAS analysis is using around 125,000 arrayed-subjects and 125,000 follow-up subjects (total n = 250,000) measured for BMI [Speliotes et al., 2011], and this approach has so far identified 32 common, low risk variants. Eighteen of these were not identified until total sample sizes reached almost 250,000, meaning that it will be difficult to determine whether rs4141463 is a true association or not, given current sample sizes in autism genetic studies of just a few thousand cases.
The contrary or complimentary hypothesis to the common variant hypothesis is that the etiology of autism is mostly accounted for by rare variants or moderate or high risk. However, since autism is under strong negative genetic selection pressure, with fecundity being about 20-fold lower than the general population [Larsen and Mouridsen, 1997], genetic risk variants of moderate or high effect should be quickly eliminated from the genetic pool through natural selection. If such variants make a significant contribution to autism, they may be of recent origin, for example, through de novo mutational events [Uher, 2009]. There is further evidence from the advanced paternal age effect seen in autism, indicative of mutations occurring in paternal gametes [Reichenberg et al., 2006], a classical mechanism for the generation of new deleterious alleles [Glaser et al., 2003], and a recent study indicated a higher rate of de novo mutation in both autism and schizophrenia compared to control subjects [Awadalla et al., 2010]. The best evidence for this, however, is empirical evidence for the role of rare, largely de novo CNVs, which at present are thought account for as much as 10% of cases. There may still be a major contribution from common risk alleles, which will be under lower selection pressure because of their small effect size (GRR < 1.3) and could theoretically escape negative selection or be balanced by positive selection from other factors. Autism may in time have much of its heritability explained through moderate risk CNVs, but hypotheses on its overall genetic architecture can only be confirmed by large-scale molecular genetic studies [Cirulli and Goldstein, 2010].
We gratefully acknowledge the following: the ongoing contribution of the parents and children in the Twins Early Development Study (TEDS), supported by a program grant (G0500079) from the UK Medical Research Council; for the German sample the families for their cooperation and professionals for collecting data, supported by grants from the Deutsche Forschungsgemeinschaft, and K. Przibilla for excellent technical assistance; Genotyping at deCODE was in part funded by EU Grant PsychCNV (HEALTH-2007-2.2.1-10-223423) and NIH grant MH071425. We also thank the kind contribution of the clinicians in the Departments of Medical Genetics and Child and Adolescent Psychiatry of the University Medical Centre Utrecht (NL). The autism research of M. Poot and W.G. Staal is supported by a grant from the Netherlands Foundation for Brain Research (Hersenstichting) (grant nos. 2008(1).34 and F2008(1)).
- 2008. Advances in autism genetics: On the threshold of a new neurobiology. Nat Rev Genet 9(5): 341–355. , .
- 2008. Case report: Autistic disorder in Kabuki syndrome. J Autism Dev Disord 38(1): 198–201. , , , .
- 2010. A genome-wide scan for common alleles affecting risk for autism. Hum Mol Genet 19(20): 4072–4082. , , , , , , , , , , et al.
- 2009. A common genetic variant in the neurexin superfamily member CNTNAP2 increases familial risk of autism. Am J Hum Genet 82(1): 160–164. , , , , , , , , , , .
- 2010. Direct measure of the de novo mutation rate in autism and schizophrenia cohorts. Am J Hum Genet 87(3): 316–324. , , , , , , , , , , et al.
- 2010. Hotspots of large rare deletions in the human genome. PLoS ONE 5(2): e9401. , , , , , , , , , , et al.
- 2007. Replicating genotype-phenotype associations. Nature 447(7145): 655–660. , , , , , , , , , , et al.
- 2008. Novel submicroscopic chromosomal abnormalities detected in autism spectrum disorder. Biol Psychiatry 63(12): 1111–1117. , , , , , , , , , , et al.
- 2009. Genomewide association studies: History, rationale, and prospects for psychiatric disorders. Am J Psychiatry 166(5): 540–556. , , , , , , , , , , et al.
- 2010. Uncovering the roles of rare variants in common disease through whole-genome sequencing. Nat Rev Genet 11(6): 415–425. , .
- 2010. Genome-wide linkage using the Social Responsiveness Scale in Utah autism pedigrees. Mol Autism 1(1): 8. , , , , , , , .
- 2010. The genetics of autism: Key issues, recent findings, and clinical implications. Psychiatr Clin North Am 33(1): 83–105. , .
- 2007. The genetics of autistic disorders and its clinical relevance: A review of the literature. Mol Psychiatry 12(1): 2–22. .
- 2009. Novel de novo SHANK3 mutation in autistic patients. Am J Med Genet Part B 150B(3): 421–424. , , , , , , , , , , et al.
- 2003. The paternal-age effect in Apert syndrome is due, in part, to the increased frequency of mutations in sperm. Am J Hum Genet 73(4): 939–947. , , , , , .
- 2009. Autism genome-wide copy number variation reveals ubiquitin and neuronal genes. Nature 459(7246): 569–573. , , , , , , , , , , et al.
- 1997. Kabuki make-up (Niikawa-Kuroki) syndrome: Cognitive abilities and autistic features. Dev Med Child Neurol 39(7): 487–490. , .
- 2008. Lack of C20orf133 and FLRT3 mutations in 43 patients with Kabuki syndrome in Japan. J Med Genet 45(7): 479–480. , , , , , , , , , , et al.
- 1997. The outcome in children with childhood autism and Asperger syndrome originally diagnosed as psychotic. A 30-year follow-up study of subjects hospitalized as children. Eur Child Adolesc Psychiatry 6(4): 181–190. , .
- 1989. Autism diagnostic observation schedule: A standardized observation of communicative and social behavior. J Autism Dev Disord 19(2): 185–212. , , , , , , .
- 1994. Autism diagnostic interview-revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. J Autism Dev Disord 24(5): 659–685. , , .
- 2009. A genome-wide association study of autism reveals a common novel risk locus at 5p14.1. Ann Hum Genet 73(Pt 3): 263–273. , , , , , , , , , , , , , , , , , , , , .
- 2007. The C20orf133 gene is disrupted in a patient with Kabuki syndrome. J Med Genet 44(9): 562–569. , , , , , , , , , , et al.
- 2010. GWAMA: Software for genome-wide association meta-analysis. BMC Bioinformatics 11: 288. , .
- 2009. Finding the missing heritability of complex diseases. Nature 461(7265): 747–753. , , , , , , , , , , , , , , , , , , , , , , , , , , .
- 2008. Structural variation of chromosomes in autism spectrum disorder. Am J Hum Genet 82(2): 477–488. , , , , , , , , , , et al.
- 2006. The discriminative ability and diagnostic utility of the ADOS-G, ADI-R, and GARS for children in a clinical setting. Autism 10(6): 533–549. , .
- 2010. Consensus statement: Chromosomal microarray is a first-tier clinical diagnostic test for individuals with developmental disabilities or congenital anomalies. Am J Hum Genet 86(5): 749–764. , , , , , , , , , , et al.
- 2009. How many genetic variants remain to be discovered? PLoS ONE 4(12): e7969. , , .
- 2011. Genetic risk sum score comprised of common polygenic variation is associated with body mass index. Hum Genet 129(2): 221–230. , , , , , , , , .
- 2003. Genetic power calculator: Design of linkage and association genetic mapping studies of complex traits. Bioinformatics 19(1): 149–150. , , .
- 2010. Genetics of osteoporosis. Ann NY Acad Sci 1192: 181–189. .
- 2003. Nonreplication in genetic association studies of obesity and diabetes research. J Nutr 133(11): 3323–3326. , .
- 2006. Advancing paternal age and autism. Arch Gen Psychiatry 63(9): 1026–1032. , , , , , , , , , . and others.
- 1996. The future of genetic studies of complex human diseases. Science 273(5281): 1516–1517. , .
- 2010. A genome-wide association study of social and non-social autistic-like traits in the general population using pooled DNA, 500K SNP microarrays and both community and diagnosed autism replication samples. Behav Genet 40(1): 31–45. , , , , , , .
- 2007. Strong association of de novo copy number mutations with autism. Science 316(5823): 445–449. , , , , , , , , , , et al.
- 2007. Rethinking the nature of genetic vulnerability to autistic spectrum disorders. Trends Genet 23(8): 387–395. .
- 2011. Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat Genet 42(11): 937–948. , , , , , , , , , , et al.
- 2007. Mapping autism risk loci using genetic linkage and chromosomal rearrangements. Nat Genet 39(3): 319–328. , , , , , , , , , , et al.
- 2009. The role of genetic variation in the causation of mental illness: An evolution-informed framework. Mol Psychiatry 12(12): 1072–1082. .
- 2009. Common genetic variants on 5p14.1 associate with autism spectrum disorders. Nature 459(7246): 528–533. , , , , , , , , , , et al.
- 2009. A genome-wide linkage and association scan reveals novel loci for autism. Nature 461(7265): 802–808. , , , .
- 2009. Elucidating the genetic architecture of familial schizophrenia using rare copy number variant and linkage scans. Proc Natl Acad Sci USA 106(39): 16746–16751. , , , , , , , .