A High-Density SNP Genome-Wide Linkage Scan in a Large Autism

Molecular Psychiatry (2009) 14, 590–600 & 2009 Nature Publishing Group All rights reserved 1359-4184/09 $32.00 www.nature.com/mp ORIGINAL ARTICLE A high-density SNP genome-wide linkage scan in a large autism extended pedigree K Allen-Brady1, J Miller1, N Matsunami2, J Stevens1, H Block1, M Farley1, L Krasny1, C Pingree1, J Lainhart1, M Leppert1, WM McMahon1 and H Coon1 1Department of Psychiatry, University of Utah, Salt Lake City, UT, USA and 2Department of Human Genetics, University of Utah, Salt Lake City, UT, USA

We performed a high-density, single nucleotide polymorphism (SNP), genome-wide scan on a six-generation pedigree from Utah with seven affected males, diagnosed with autism spectrum disorder. Using a two-stage linkage design, we first performed a nonparametric analysis on the entire genome using a 10K SNP chip to identify potential regions of interest. To confirm potentially interesting regions, we eliminated SNPs in high linkage disequilibrium (LD) using a principal components analysis (PCA) method and repeated the linkage results. Three regions met genome-wide significance criteria after controlling for LD: 3q13.2–q13.31 (nonparametric linkage (NPL), 5.58), 3q26.31–q27.3 (NPL, 4.85) and 20q11.21–q13.12 (NPL, 5.56). Two regions met suggestive criteria for significance 7p14.1–p11.22 (NPL, 3.18) and 9p24.3 (NPL, 3.44). All five chromosomal regions are consistent with other published findings. Haplotype sharing results showed that five of the affected subjects shared more than a single chromosomal region of interest with other affected subjects. Although no common autism susceptibility genes were found for all seven autism cases, these results suggest that multiple genetic loci within these regions may contribute to the autism phenotype in this family, and further follow- up of these chromosomal regions is warranted. Molecular Psychiatry (2009) 14, 590–600; doi:10.1038/mp.2008.14; published online 19 February 2008 Keywords: autism spectrum disorder; genome-wide scan; chromosome 3; chromosome 20; chromosome 7; chromosome 9

Introduction of such genomic searches have been published to date.5–17 However, no single chromosomal region has Autism (MIM 209850) is a pervasive neurodevelop- been consistently shown to be associated with autism. mental disorder, characterized by impairments in Numerous reasons for the lack of consistency of verbal and nonverbal communication and social results exist including genetic heterogeneity, such interaction, and the presence of repetitive stereotyped that multiple genes are thought to be involved in the behaviors and interests. Autism is a member of a etiology of autism.9,18 It has been suggested that as few spectrum of neurodevelopmental disorders (autism as 3–4 genes19 and perhaps up to 15 predisposing spectrum disorders (ASDs)), which also includes genes9 may be involved, and the complexity of Asperger’s syndrome, and pervasive developmental finding multiple genes is further increased as autism disorder, not otherwise specified (PDD-NOS). It is predisposition genes do not appear to be inherited in typically diagnosed within the first 3 years of life, and a simple Mendelian fashion. Furthermore, genetic there is strong evidence that autism is highly and environmental interactions may also contribute heritable. The concordance rate in monozygotic twins to the autism phenotype.20,21 In addition, the inclusion is 70–90%,1,2 and the autism rate in siblings is 3–5%, of heterogeneous phenotypes may explain the lack of much higher than expected from the general popula- consistency of results in the study of autism. Genome- tion prevalence.3,4 Despite the strong heritability, wide linkage scans that stratify subjects into more identification of the underlying genetic mechanisms homogeneous phenotypes have shown striking differ- for autism has been elusive. ences in linkage results, including differences by the Efforts to find autism predisposition genes have strictness of the definition of autism,16,22 sex,16,23 focused on genome-wide linkage scans, and a number developmental regression,16,24 language delay25,26 and repetitive behaviors.27,28 The key to identification of Correspondence: Dr K Allen-Brady, Utah Autism Research autism susceptibility genes will be the reduction of both Program, 650 Komas Drive, Suite 206, Salt Lake City, UT 84108, genetic and phenotypic heterogeneity through selection USA. E-mail: [email protected] of well-characterized families. Received 30 November 2006; revised 21 December 2007; accepted While most linkage studies to date have utilized 2 January 2008; published online 19 February 2008 nuclear families composed of either trios (two parents High-density linkage in large autism pedigree K Allen-Brady et al 591 and an affected offspring), or sibling pairs (concor- Materials and methods dant or discordant), large extended pedigrees offer many advantages for the localization of candidate Subjects genes for autism. The study of large extended The seven affected male subjects are all descendants pedigrees increases the likelihood that there will be of a single founding couple of Northern European greater phenotype homogeneity and reduced environ- ancestry (see Figure 1). To preserve confidentiality of mental heterogeneity by pedigree. Study of large the pedigree, most siblings are excluded. Diagnosis pedigrees also increases the power of a study to was based on Diagnostic and Statistical Manual of detect disease predisposition genes because of re- Mental Disorders, fourth edition, Text Revision duced genetic heterogeneity by pedigree. While study (DSM-IV-TR) criteria. Developmental history and of large autism pedigrees affords many advantages, clinical observations were gathered using the Autism because of the relatively low autism recurrence risk in Diagnostic Interview-Revised (ADI-R) and the relatives, finding large extended autism pedigrees is a Autism Diagnostic Observation Schedule-Generic challenge. (ADOS-G),30,31 respectively, with two exceptions. Here, we present results for a high-density One subject was unavailable for ADOS-G testing, single nucleotide polymorphism (SNP) genome- and another subject’s ADI-R was considered unreli- wide linkage analysis in a six-generation pedigree able, as the informant was elderly and unable to recall from Utah with seven affected males, which is one subtle aspects of early development. However, in both of the largest known autism pedigrees. Previously, cases there was also documentation of an earlier we published linkage results using this same pedigree diagnosis and/or very early concerns about social for a single chromosomal region (3q25–27).29 In development such that, along with current informa- our current study, we present results for a genome- tion gathered through our testing, DSM-IV-TR criteria wide two-stage linkage design. In stage 1, an for autistic disorder was met. Six of the seven affected initial linkage screen was performed to identify subjects met DSM-IV-TR criteria for autistic disorder. potential regions of interest. We confirmed potentially One subject (subject no. 5) met DSM-IV-TR criteria for interesting regions in stage 2 by eliminating SNPs PDD-NOS. As one of the seven affected subjects was in high LD. Maximizing genetic and phenotypic diagnosed with PDD-NOS, we define the entire homogeneity through study of a highly informa- pedigree as an ASD pedigree. tive pedigree using an increased genetic-marker Other clinical characteristics of the seven affected density increases the probability of both confirming subjects have been assessed; however, little similarity and narrowing significant linkage regions between cases was observed. Full-scale intelligence previously published for autism, as well as further quotient (IQ) scores of affected subjects ranged from facilitating the identification of autism predisposition 41 to 124 for verbal IQ (VIQ) and 45 to 140 for genes. performance IQ (PIQ) using IQ measures appropriate

****

* * * * * * * * * * 6

* * * * 1 3 * 4 * *

2 * 7 5 Figure 1 Basic pedigree structure of extended autism pedigree. Affected subjects are shaded and numbered; deceased subjects are indicated with a slash. All unaffected persons shown in this figure with an asterisk (*) were genotyped as were all affected subjects (7 affected subjects and 22 relatives).

Molecular Psychiatry High-density linkage in large autism pedigree K Allen-Brady et al 592 for age and level of functioning (Wechsler Adult SNP genotyping Intelligence Scale (WAIS-III),32 Wechsler Intelligence We genotyped the 7 affected individuals and the 22 Scale for Children (WISC-III)33 or Differential Abili- unaffected relatives using the Affymetrix 10K SNP ties Scale (DAS)34). Three of the seven affected panel. The 10K SNP panel contains 10 660 SNPs on a subjects showed language delay (that is, onset of single array, and has an average distance between single words after 24 months and/or onset of phrases SNPs of 210 kb. SNP genotypes were obtained by after 33 months) as measured by items on the ADI-R. following the Affymetrix protocol for the GeneChip One of the affected subjects had a history of Mapping 10K Xba Array.37 In brief, 250 ng of genomic nonfebrile seizures, but the other six did not. None DNA taken from peripheral blood was digested with of the seven subjects showed developmental regres- the Xba1 restriction enzyme into fragments. This step sion (that is, loss of previously gained language and is followed by ligation to Xba adaptors that place no social skills) as measured by items on the ADI-R. See restriction on fragment size. Using PCR reaction, a Table 1 for details of these phenotypes. single generic primer is used to amplify adaptor- We included 22 unaffected living relatives from the ligated DNA fragments. The amplified DNA product same pedigree to infer genotypes of deceased relatives is then fragmented, labeled with biotin-ddATP and or phase for genotyped cases. Phenotypes for the hybridized to the 10K chip. Following an 18-hour unaffected relatives were set to ‘unknown’ or missing hybridization, the chip is washed, stained and for all analyses, as phenotype status of the deceased scanned using an Affymetrix Fluidics Station FS450 relatives was not recorded or possible to determine. and the Affymetrix GeneChip 3000 scanner. Affyme- Records were also not available in sibships of upper trix GCOS software was used to determine SNP generations of the pedigree; it is unknown if there genotypes for each locus. were additional affected members and hence antici- pation could not be assessed. We verified pedigree Genotype error checking and mapping structure for the 7 affected and 22 unaffected relatives Genotype error checking including checks for Men- using the Utah Population Database (UPDB), a delian inconsistencies was performed with Mega2,38 computerized genealogy database that contains family and as a second validation we used the program history information for almost 4 million individuals CheckErrors,39 which uses graphical modeling to who are, for the most part, descendants of the calculate the posterior probability of genotype errors nineteenth century pioneers to Utah. Low rates of in pedigrees. All SNPs with genotype errors from inbreeding have been previously reported within the either Mega2 or CheckErrors were eliminated from UPDB.35,36 Fragile X DNA testing was performed on further study. We also eliminated all SNPs with a rare all known autism cases and found to be negative. allele frequency p0.10. Karyotyping using a routine g-banded chromosomal We used the genetic map provided by Affymetrix, analysis to rule out major chromosomal rearrange- based on the deCODE genetic map, for the linkage ments was also performed on all cases; no large-scale analysis. Base pair positions were obtained from the chromosomal abnormalities were observed. All sub- May 2004 human reference sequence (hg17) assembly jects signed an informed consent and this study was and were used for graphs presented in the manu- approved by the University of Utah Institutional script. For peak regions that spanned the centromere, Review Board. we validated separately the chromosomal p and q

Table 1 Phenotypic characteristics of the seven affected male subjects in the pedigree

Subjecta Age VIQ PIQ ADI ADI ADI ADOS-Gb Regressive Language Seizures (years) social communication repetitive onset delay

1 48 124 140 25 9 4 18 No No No 2c 28 41 45 24 24 7 —c No Yes Yes 327488526231222NoNoNo 433859326181212NoNoNo 512—d —d 7 4 4 7 No No No 6438886—e —e —e 11 No Yes No 7 29 45 45 24 16 3 15 No Yes No

Abbreviations: ADI-R, Autism Diagnostic Interview-Revised; ADOS-G, Autism Diagnostic Observation Schedule-Generic; PIQ, performance intelligence quotient; VIQ, verbal intelligence quotient. aSubjects were numbered arbitrarily. bCombined communication and social algorithm score. cSubject 2 is not available for further ADOS-G testing. dInformation is not available; however, Standford Achievement Test Scores in the fourth grade suggest that the subject’s IQ is at least 70 eADI for subject 6 is not considered reliable and another close informant is not available.

Molecular Psychiatry High-density linkage in large autism pedigree K Allen-Brady et al 593 arms for the LD assessment, because of suppressed The D0 threshold value of 0.70 has been used by recombination in the centromere region. others to eliminate SNPs in high LD.46,47 Markers identified as being in high LD were phased using the Stage 1: initial genome-wide linkage screen software SNPHAP, and the resulting haplotypes were An initial genome-wide linkage scan with no correc- entered into a principal components analysis (PCA) tion for LD was performed using the multipoint in a manner using a modified version of the PCA tag- linkage software MCLINK.40 MCLINK is a Markov SNP method for candidate genes proposed by Horne chain Monte Carlo method that allows for fully and Camp.48 In brief, this PCA method extracts factors informative multilocus linkage analysis on large that account for at least 90% of the underlying genetic extended pedigrees. Using blocked Gibbs sampling, variation; each extracted factor is defined as an LD MCLINK generates inheritance matrices from haplo- group. For each LD group, we retained the SNP with the type chains for the markers being analyzed, and highest factor loading as the SNP that best characterizes performs an approximate calculation of the log-like- the variance for each group, and eliminated all other lihood function linkage statistics. MCLINK has been SNPs as providing redundant information. Hence, SNPs used previously to identify candidate genomic re- that were included in the confirmation analysis (1) gions for a number of complex diseases.41–44 We were included in HapMap data and (2) had either performed a nonparametric linkage analysis, as D0 values of < 0.7 or if D0 was X0.7, the SNP was inheritance models for autism in general and speci- selected from the PCA method. Linkage analysis was fically for broad spectrum of autism are unknown. repeated using a reduced marker list following the same Ideally, we could have derived the inheritance model procedure as described previously. parameters from our sample and used them for analysis. However, because we have a single pedigree, segregation parameters could not be resolved with Significance any degree of accuracy and hence a model-free As nonparametric linkage results are on a different approach was utilized. Allele frequencies for the scale than parametric LOD scores, we applied the MCLINK analysis were estimated using a maximum NPL equivalent of the Lander and Kruglyak criteria49 likelihood gene counting method,45 which statisti- for determining suggestive and significant results (see cally infers allele frequencies based on the assump- also Ray and Weeks50). For the initial LD screening tion that the pedigree founder alleles were randomly analysis, a potentially interesting region was defined sampled from a wider population. For this study, as having an NPL score X2.00; LD assessment was allele frequencies as defined by Affymetrix for performed for all of these regions. After elimination of their 10K SNP panel on 40 Caucasian subjects SNPs in high LD, we applied the Lander and Kruglyak were considered as the wider population allele significance criteria to determine significance. Sug- frequencies. gestive evidence of linkage was defined as an NPL score X3.18, (P = 0.00074) and significant evidence Stage 2: assessment of linkage disequilibrium in of linkage was defined as an NPL score X4.08 regions with linkage NPL results X2.0 (P = 0.000022). Boundary regions for a peak after LD Currently available multipoint linkage software assessment were defined based on a ‘two NPL drop’ assumes that genetic markers are in linkage interval. For regions of interest at the beginning or equilibrium, which is a valid assumption if one uses end of the chromosomal p or q arms, where it was not distantly spaced microsatellite markers. However, feasible to require that a boundary SNP reach a ‘two using higher density, more closely spaced SNP NPL drop’ criteria, the first or last available SNP in markers may violate this assumption. For all regions the area defined the region of interest. As a secondary from stage 1 with nonparametric linkage (NPL) scores validation of significance, we also simulated 100 null X2.0 and an additional B15–20 SNPs on either side genotype configurations using the software package of a peak region, we screened for potential biased SIMULATE51 for all potentially interesting regions results due to LD. As the number of unrelated (NPLX2.0). SIMULATE51 generates genotype data for individuals in our pedigree was limited, we used linked markers unlinked to affection status using the unrelated, parental genotype data (N = 60) from Centre same pedigree structure, population allele frequen- d’Etude du Polymorphisme Humain (CEPH) Utah cies and recombination fractions between markers as trios in HapMap for LD assessment. The CEPH the original data. As analysis of SNP data can lead to founders, as also being of Northern and Western biased linkage results because of LD between markers, European ancestry, are likely to be very similar to our only regions where we had previously removed high- Utah autism extended pedigree subjects, and are a LD markers were reanalyzed. For each reanalyzed valid resource for determining LD structure for our region, we performed a nonparametric linkage analy- subjects. Using only Affymetrix SNPs that matched sis using MCLINK and compared the maximum NPL HapMap CEPH genotype data, we estimated the result from the simulated null genotype data to the pairwise LD measure D0 between all pairs of maximum NPL result obtained from the observed data SNPs in a region of interest. All markers with a for the region. The reported P-value is a count of D0 value X0.70 and within 2-million base pairs the number of times the simulated data exceeded the of each other were considered for potential removal. observed data across the 100 simulations.

Molecular Psychiatry High-density linkage in large autism pedigree K Allen-Brady et al 594 Results chromosome 20 were all shared among four affected subjects. The region on chromosome 7 was shared by Data quality three affected subjects. Two distinct haplotypes were All 7 affected individuals and the 22 unaffected shared by two different pairs of affected subjects on relatives were genotyped on the premarket version of chromosome 9. It should be noted that the groups the Affymetrix 10K chip. Because of sample errors, of individuals who shared a particular haplotype two subjects were re-genotyped on a subsequent differed by chromosomal region, such that the version of the 10K chip. From the 10 660 total SNPs group of individuals, for example, who shared the on a 10K chip, we eliminated 407 SNPs because of chromosome 3q13.2–q13.31 region differed from the differences in chip versions and 412 SNPs because of group of individuals who shared the chromosome obvious mapping errors. We eliminated an additional 3q26.31–q27.3 region. Individual 2 did not share any 1634 SNPs because the minor allele frequency was of the chromosomal regions of interest based on 0.10 and 195 SNPs because of genotype errors. The p linkage evidence with his other affected relatives, final total number of SNPs studied was 8012. whereas individual 6 shared all of the chromosomal Linkage analyses regions of interest with other affected relatives. The results from the initial multipoint genome-wide linkage scan obtained from MCLINK are displayed in Discussion Table 2 and Figure 2. There were 18 regions with an NPL score X2.00. The maximum NPL statistic We present results for a high-density, genome-wide was 5.56 (empirical P-value 0.000000287) and was SNP analysis in a single large pedigree with seven found on both chromosome 3p12.3–q21.3 and affected autism cases from Utah. Our six-generation 20p12.1–q13.12. ASD pedigree of all male cases represents one of the largest autism pedigrees known. Through study of this highly informative pedigree where cases are more LD assessment results likely to be genetically homogeneous, we have the Linkage disequilibrium assessment was performed for potential to confirm previously published linkage the 18 chromosomal regions attaining an initial NPL results as well as narrow significant linkage regions score X2. 0. From the original set of Affymetrix 10K with an end goal of facilitating the identification of SNPs for each region of interest, we retained on autism susceptibility loci. average 87.3% (s.d. 4.8%) of SNPs after matching to The most promising chromosomal regions to HapMap data and on average 69.8% (s.d. 8.6%) of emerge from these results are the 3q13.2–q13.31, SNPs after matching to HapMap data and after 3q26.31–q27.3 and the 20q11.21–q13.12 regions. The elimination of SNPs in high LD. Table 2 shows the maximum multipoint NPL score was obtained for median intermarker distance for chromosomal re- chromosome 3q13.2–q13.31 (NPL, 5.58; empirical gions of interest both prior to LD assessment and after P = 0.000000287). Schellenberg et al.16 also observed removal of SNPs in high LD. The overall median a significant peak in the 3q13.31 region using intermarker distance for all regions of interest after microsatellite markers for families with two affected matching HapMap data, but prior to LD assessment siblings. Their most significant finding (P = 0.007) for was 0.248 Mb (25th and 75th percentile 0.091– the 3q13.31 region was for families without behavior- 0.565 Mb). After elimination of high-LD SNPs, the al regression and a broad diagnosis of autism, which overall median intermarker distance was 0.344 Mb included 147 families where both affected siblings (25th and 75th percentile 0.156–0.709 Mb). met criteria for the strict definition of autism and 53 The validation linkage results after removal of SNPs families with other types of affected sibpairs (for in high LD are presented in Table 2 and Figure 3. Five example, a sibpair diagnosed with strict definition chromosomal regions had NPL scores X3.18 after LD of autism and the second sibling diagnosed with assessment. The maximum multipoint NPL score of PDD-NOS). Other subset analyses of their data were 5.58 (empirical P = 0.000000287) was obtained for also found to be significant at the same 3q13.31 chromosome 3q13.2–q13.31. Other regions attaining region: multipoint (P = 0.03) and single point significant NPL scores (NPLX4.08) were 3q26.31– (P = 0.02) analyses for families with a broad diagnosis q27.3and 20q11.21–q13.12. Regions attaining sugges- of autism without consideration of behavioral regres- tive evidence for linkage (NPLX3.18) were 7p14.1– sion, families with only male affected siblings p11.22 and 9p24.3. We note that the simulation (P = 0.03) and families without behavioral regression generated results concurred with the observed data. and a strict diagnosis of autism (P = 0.02). These For all five regions of interest meeting at least results also concurred with a meta-analysis of five suggestive evidence of linkage, the observed NPL genome scans for a broad definition of autism- result exceeded all 100 simulation-generated NPL spectrum disorders that found interesting results for results (that is, P < 0.01). the 3q13.1–q21 region (P = 0.021).22 Based on our Table 3 contains the haplotype sharing information current results and these two other studies, genetic for the five chromosomal regions attaining at least variant(s) in the 3q13.2–q13.31 region may predispose suggestive evidence of linkage (NPLX3.18). The two to broad spectrum of autism in males without regions of interest on chromosome 3 and the region on behavioral regression. As further corroborating

Molecular Psychiatry Table 2 Results of linkage screening and linkage validation analyses

Linkage screening analyses Linkage validation analyses after removal of high-LD SNPs

Chromosomal Max NPL Empirical Distance Median intermarker Median intermarker Chromosomal Region of Max NPLe Empirical Simulated regiona (P-valueb) studied (Mb) distance prior to LD distance after LD regiona interestd (P-valueb) (P-valuef) removal (Mb)c removal (Mb)c

1p13.2–p13.1 2.48 0.00657 11.69 0.141 0.242 1p13.2 2.34 0.00964 0.05 2q32.2–q32.3 2.32 0.0102 27.69 0.221 0.284 2q32.2–q32.3 2.31 0.0106 0.02 3p14.1 2.45 0.00724 18.24 0.420 0.424 3p14.1 2.44 0.00742 < 0.01 3p12.3–q21.3c 5.56 0.000000287 18.99 0.416 0.589 3p12.3–p11.1 2.40 0.0083 0.03 45.45 0.406 0.493 3q11.2–q21.3 3q13.2–q13.31 5.58 0.000000287 < 0.01 3q24–q25.1 2.70 0.00351 19.81 0.317 0.346 3q24–q25.1 2.85 0.00224 < 0.01 3q26.31–q27.2 5.45 0.000000287 43.42 0.347 0.412 3q26.31–q27.2 3q26.31–q27.3 4.85 0.000000636 < 0.01 4p16.3–p15.33 2.44 0.00742 16.47 0.200 0.293 4p16.3–p15.33 2.4 0.00827 0.03 7p14.3–p11.2g 5.54 0.000000287 38.07 0.185 0.314 7p14.3–p14.1 2.88 0.00203 0.01 7p14.1–p11.1 7p14.1–p11.22 3.18 0.000752 < 0.01 7q21.11 2.52 0.00593 22.47 0.297 0.457 — — — 0.52 9p24.3 3.87 0.0000551 8.81 0.111 0.231 9p24.3 9q24.3 3.44 0.000291 < 0.01 10q25.1 2.17 0.0153 15.04 0.230 0.323 — — — 0.08 11q22.2–q22.3 2.44 0.00742 12.77 0.098 0.271 11q22.2–q22.3 2.44 0.00742 0.02 11q24.2 2.13 0.0167 10.22 0.130 0.231 11q24.2 2.02 0.0218 0.07 15q14–q15.1 2.43 0.00766 26.25 0.369 0.443 — — — 0.01 17q23.3 2.46 0.007 20.26 0.241 0.384 17q23.3 2.46 0.00702 0.02 ihdniylnaei ag uimpedigree Allen-Brady autism K large in linkage High-density 20p12.1–q13.12h 5.56 0.000000287 13.95 0.148 0.240 20p12.1–p11.21 2.64 0.00418 0.01 23.13 0.321 0.506 20q11.21–q13.12 20q11.21–q13.12 5.56 0.000000287 < 0.01 22q11.1–q11.21 3.82 0.0000677 23.22 0.303 0.480 — — — 0.09

22q12.3 2.1 0.0178 23.22 0.303 0.480 — — — 0.07 al et

Abbreviations: LD, linkage disequilibrium; Max, maximum; NPL, nonparametric linkage; SNPs, single nucleotide polymorphism. aLocation of maximum NPL scores > 2.0. bEmpirical P-value determined by MCLINK. cSNPs not included in HapMap CEPH Utah data were excluded prior to calculation of median intermarker distance. dRegion defined based on a ‘two NPL drop’, except for genomic regions at the beginning or end of the chromosomal p and q arms meeting threshold criteria where the first available marker defines the boundary. eValues that are bold faced and underlined indicate significant linkage evidence based on Lander and Kruglyak (1995) criteria. Bold only values indicate suggestive linkage evidence. fSimulation P-value determined using 100 Monte Carlo simulations. gPeak region spans centromere. Because of suppressed recombination in the centromere region, the chromosomal p and q arm were validated separately. hLinkage validation analyses resulted in two separate and distinct peaks. oeua Psychiatry Molecular 595 High-density linkage in large autism pedigree K Allen-Brady et al 596

5 4 3 2 1 0 -1

1 2 3 4 5 6 7 8

5 4 3 2 1 0 -1 910111213141516171819202122X Figure 2 Nonparametric linkage (NPL) scores across the genome prior to linkage disequilibrium (LD) assessment.

evidence for the significance of the 3q13.1–q21, we specific for broad spectrum of autism. Further, as also note that the Autism Genome Project Consortium increased linkage evidence for the chromosome 3 recently published a copy number variant analysis in region was not observed in a combined analysis of the which they detected a copy number variant loss at same Finnish families and the large-scale collabora- chromosome 3q13.31 that was validated by familial tive efforts of the Autism Genetic Resource Ex- clustering.17 Autism candidate genes of interest in the change,54 this suggests that this particular 3q13.2–q13.31 region include DRD3, GAP43, LSAMP mutation(s) may be limited to those with Northern and MAK3. In an early publication, allele frequencies European ancestry. We make particular note that our in the gene DRD3 were compared between subjects current study using high-density SNPs was able to with autism and matched school children; however, narrow the region from 3q25–27 identified using no significant difference between groups was ob- microsatellite markers in the Finnish study to served.52 3q26.31–q27.3. We also point out that in our previous The second highest peak was observed at study,29 we did not attempt to rigorously control for 20q11.21–q13.12 (NPL, 5.56; empirical P = 0.000000287). LD; hence, we reported our previous findings as Schellenberg et al.16 also found suggestive linkage within the general 3q25–27 region. Potential candi- results at 20q11.21; however, only for the subset of date genes for autism in this chromosomal region families with male only affected siblings and a broad include NLGN1, FXR1, SOX2, HTR3C, HTR3D, diagnosis of autism as defined above (P = 0.03). Again, HTR3E, AHSG, CAM-KIIN and EPHB3. FXR1 and based on our current results and those of Schellenberg NLGN1 were previously screened as potential autism et al.,16 genetic variants in the 20q11.21–q13.12 region candidate genes; however, neither showed a strong may predispose males to broad definition of autism. association with autism.29,55,56 Autism candidate genes of interest in this region For the two chromosomal regions with suggestive include NNAT, SLC32A1, POFUT1, DLGAP4 and evidence of linkage, 7p14.1–p11.22 and 9p24.3, both PTPRT. However, none of these genes have been were replicated by other groups. Wolpert et al.57 tested for an association with autism. observed a single case with autism to have a de novo The 3q26.31–q27.3 region (NPL, 4.85; empirical partial duplication of 7p11.2–p14.1. As noted by P = 0.000000636) was previously identified by our Wolpert et al.,57 the 7p11.2–p14.1 region is adjacent to group,29 and replicates findings for 38 Finnish autism the HoxA-1 gene, which has been suggested as a families to 3q25–27 (maximum multipoint LOD candidate gene for autism.58–60 A copy number variant score = 4.81), particularly their follow-up study of 34 gain was detected by the Autism Genome Project families from Central Finland.53 The strongest results Consortium at chromosome 9p24.3.17 for the 3q25–27 region from the Finnish study were Based on our haplotype sharing results for the five obtained for a broad autism spectrum diagnosis that chromosomal regions with at least suggestive evi- included Asperger’s syndrome. Our study of an ASD dence of linkage results (NPLX3.18), no single pedigree of Northern European descent confirms the haplotype was consistently shared among all affected Finnish results, and suggests that this locus might be subjects, suggesting that no single chromosomal

Molecular Psychiatry High-density linkage in large autism pedigree K Allen-Brady et al 597 Chr 3q13.2-q13.31 Chr 3q26.31-q27.3

6 Auranen et al. 2002 rs262980 max NPL=4.85 Broad diagnosis 5 LOD = 4.81 rs1432388 6 max NPL=5.58 Schellenberg et al. 2006 4 5 Regression negative/ broad diagnosis 4 p=0.007 3 NPL 3 rs712530 rs1402623 rs1882256

NPL 2 2 1 rs1365111 1 0 0

.6 .4 .2 .1 .9 .4 7 .1 0 9 .0 0 6 5 9 .5 .3 .9 9 .9 .0 .6 2 0.6 4.5 7.7 9.4 2.5 4 7 8 9 4 7. 2 4. 8.2 6. 1. 4. 6.7 7. 9 1 3 5. 8 9.9 2 3 4. 95 0 1 2 3 3 6 72. 7 7 8 8 8 9 9 1 10 10 10 11 11 11 1 11 12 1 1 1 13 165.41 169 17 1 1 17 1 17 18 1 1 18 1 1 19 1 position (Mb) position (Mb)

Chr 7p14.1-p11.22 Chr 9q24.3

rs726873 3.5 max NPL=3.18

3 rs1576748 rs588412 4 maxNPL=3.44 2.5 3.5 2 3 2.5 NPL 1.5 rs1112534 rs720816 2 NPL 1 1.5 Wolpert et al. 2001 rs2222544 1 0.5 inverted duplication of 7p11.2-p14.1 0.5 in a single case with AD 0 0

5 .6 .5 .0 .4 .5 .9 .5 .6 .7 0.2 0.3 0.5 0.6 1.5 2.0 2.1 2.1 2.2 2.3 2.9 3.0 3.1 3.6 4.1 5.3 8 9 2 4 1 7 9 41.5 42.4 42.9 45.1 46. 47.3 4 4 50 51 5 5 6 6 6 position (Mb) position (Mb)

Chr 20q11.21-q13.12

rs721220 6 5 Schellenberg et al. 2006 rs718092 max NPL=5.56 4 Male-only/ broad diagnosis 3 p=0.03

NPL rs988462 2 1 0

.5 .2 .9 .8 .7 .3 .9 .1 .4 .1 .8 .6 9 4 5 7.7 8 9 0 0 2 4 5 7 8 2 32.3 3 3 3 3 3 4 4 4 4 4 4 4 50.1 position (Mb)

Figure 3 Plots of the linkage results after removal of high LD SNPs for regions of the genome showing at least suggestive evidence of linkage. (a) Chr3q13.2–q13.31; (b) Chr3q26.31–q27.3; (c) Chr7p14.1–p11.22; (d) Chr9q24.3; (e) Chr20q11.21–q13.12.

Table 3 Haplotype sharing for seven affected subjects across five chromosomal regions with at least suggestive evidence of linkage for autism

Subject 3q13.2–q13.31 3q26.31–q27.3 7p14.1–p11.22 9p24.3a 20q11.21–q13.12

1XXAX 2 3XX BX 4X B 5XX X 6XX XAX 7 X aTwo distinct haplotypes (labeled ‘A’ and ‘B’) were shared among two pairs of individuals for the chromosome 9p24.3 region. region, or hence no single gene, explained all of the 6) shared more than a single chromosomal region autism cases observed. Rather, we observed that five of interest with other affected subjects. For these of the affected subjects (that is, subjects 1, 3, 4, 5 and five affected subjects, sharing occurred across a

Molecular Psychiatry High-density linkage in large autism pedigree K Allen-Brady et al 598 combination of various different regions, via most for phenotype classification, particularly in pedigrees likely gene–gene interactions, to result in the autism where there is only a single child in a nuclear family phenotype. The hypothesis of genetic heterogeneity with ASD. in autism is not new.16,61 However, what is interesting In our analysis, we controlled for the presence of is that we observed genetic heterogeneity within a LD among SNP markers, as LD has the potential to single large pedigree. While it is possible that more bias multipoint linkage results. The majority of than a single autism predisposition gene may lie in previous studies that have examined the effect of LD one of the identified regions, our findings suggest that on linkage analyses have found an inflation of both the number of potential genes and various combina- NPL and LOD scores,46,63,64 although a single study tions of genes that may be involved in autism within observed a slight decrease in NPL scores in the our pedigree are limited, and we have identified a presence of LD.65 We observed 18 chromosomal relatively small number of regions to follow-up. regions attaining an NPL score X2.0 when LD was Furthermore, genetic heterogeneity within a single ignored, and 16 of these regions resulted in NPL pedigree may also help explain the lack of consis- scores X2.0 when LD was considered. We found that tency of results between autism studies. If genetic our methodology for eliminating high-LD SNPs was heterogeneity exists in a single large pedigree, it consistent and actually more conservative than that likely exists in other autism pedigrees and further used by Sellick et al.,64 who also used a 10K SNP emphasizes the challenges that will be faced to find panel for study of a complex disease. We observed a autism predisposition genes. Despite the genetic median intermarker distance of 0.344 Mb (25th and heterogeneity observed, the reduced phenotypic and 75th percentile 0.156–0.709 Mb) after elimination of environmental heterogeneity in our large pedigree high-LD SNPs, and Sellick et al.64 observed a median allowed us to detect three significant and two intermarker distance of 0.17 Mb (25th–75th percentile suggestive linkage peaks. 0.06–0.38) after removal of high-LD SNPs by selecting We did not observe haplotype sharing for indivi- a single SNP from clusters of SNPs with an r2 > 0.4. It dual 2 with the other affected subjects at any of the is possible that LD may still have influenced our five chromosomal regions of interest. Individual 2 results, and it is also possible that we may have may be a sporadic autism case in the pedigree with missed other potentially interesting regions by not de novo mutations and/or environmental causal assessing LD on the complete genome. Our high-LD factors, as he had the lowest IQ and is the only case elimination method is rather time intensive, and to have seizures. It is also possible that this family is hence we opted to assess LD only in regions of genetically heterogeneous and despite the relation- potential interest defined at a conservative threshold. ship between affected individuals, the occurrence of As all five chromosomal regions with at least autism in some of the branches of the genealogical suggestive evidence of linkage are consistent with tree might have appeared independently. Lack of other linkage analyses using microsatellite markers or sharing with individuals 2 could also indicate that analyses with copy number variants, we feel con- not all genes contributing to autism within this fident that these regions are valid peaks and warrant pedigree were detected. Here, we focused on results follow-up with fine mapping and candidate gene attaining at least suggestive evidence of linkage. As studies. shown in Table 2 there are other regions of the In conclusion, our genome-wide linkage analysis of genome that are potentially interesting but have NPL a single high-risk ASD pedigree resulted in suggestive scores less than 3.18. Furthermore, submicroscopic or significant linkage evidence for five distinct chromosomal abnormalities may also predispose to chromosomal regions, all of which show striking autism. A follow-up study to estimate copy number replication with other autism studies. Our replication variants using higher density SNPs is planned. While of previous results adds credibility to previous it is possible that multiple genes across these other findings and reinforces the soundness of our metho- regions may act jointly in a low-penetrant polygenic dology for identifying regions of interest, and follow- manner to contribute to the autism phenotype, up of these five chromosomal regions is warranted. additional studies are required to understand their possible contribution to autism. Acknowledgments One of the most consistent chromosomal regions demonstrating significant linkage results for the This work was supported by R01 MH069359, 5 U19 autism phenotype is at chromosome 7q,9,10,13,14,16,62 a HD035476 (one of the NICHD Collaborative Programs region that was not significant in our analyses. Recent of Excellence in Autism), the Utah Autism Founda- studies have shown that the peak on chromosome 7q tion and by GCRC grant number M01-RR00064 from is only significant in studies using subjects with a the National Center for Research Resources. Partial strict diagnosis of autism.16,22 When the diagnosis is support for all datasets within the Utah Population relaxed and ASDs are included in the phenotype, the Database (UPDB) was provided by the University of linkage evidence is much weaker.16,22 Despite only Utah Huntsman Cancer Institute. We thank Dr Sally one of seven cases being diagnosed with PDD-NOS in Ozonoff for assistance with diagnoses of subjects, and our extended pedigree, our negative results at chro- our staff whose countless hours of work have made mosome 7q emphasize the importance of genealogy this study possible. We also greatly appreciate the

Molecular Psychiatry High-density linkage in large autism pedigree K Allen-Brady et al 599 time and effort given by the family members who 18 Folstein SE, Rosen-Sheidley B. Genetics of autism: complex participated in this study. aetiology for a heterogeneous disorder. Nat Rev Genet 2001; 2: Electronic-Database Information 943–955. 19 Pickles A, Bolton P, Macdonald H, Bailey A, Le Couteur A, Sim CH Online Mendelian Inheritance in Man (OMIM), et al. Latent-class analysis of recurrence risks for complex http://www.ncbi.nlm.nih.gov/Omim/ phenotypes with selection and measurement error: a twin Utah Population Database (UPDB), http://www.hci. and family history study of autism. Am J Hum Genet 1995; 57: 717–726. utah.edu/groups/ppr/ 20 Jones MB, Szatmari P. A risk-factor model of epistatic interaction, HapMap, http://www.hapmap.org/ focusing on autism. Am J Med Genet 2002; 114: 558–565. SNPHAP, http://www-gene.cimr.cam.ac.uk/clayton/ 21 Ma DQ, Whitehead PL, Menold MM, Martin ER, Ashley-Koch AE, Mei H et al. Identification of significant association and gene–gene software/ interaction of GABA receptor subunit genes in autism. Am J Hum Genet 2005; 77: 377–388. 22 Trikalinos TA, Karvouni A, Zintzaras E, Ylisaukko-oja T, References Peltonen L, Jarvela I et al. A heterogeneity-based genome search meta-analysis for autism-spectrum disorders. Mol Psychiatry 2006; 1 Bailey A, Le Couteur A, Gottesman I, Bolton P, Simonoff E, Yuzda E 11: 29–36. et al. Autism as a strongly genetic disorder: evidence from a British 23 Stone JL, Merriman B, Cantor RM, Yonan AL, Gilliam TC, twin study. Psychol Med 1995; 25: 63–77. Geschwind DH et al. Evidence for sex-specific risk alleles in 2 Steffenburg S, Gillberg C, Hellgren L, Andersson L, Gillberg IC, autism spectrum disorder. Am J Hum Genet 2004; 75: 1117–1123. Jakobsson G et al. A twin study of autism in Denmark, Finland, 24 Molloy CA, Keddache M, Martin LJ. Evidence for linkage on 21q Iceland, Norway and Sweden. J Child Psychol Psychiatry 1989; 30: and 7q in a subset of autism characterized by developmental 405–416. regression. Mol Psychiatry 2005; 10: 741–746. 3 Bolton P, MacDonald H, Pickles A, Rios P, Goode S, Crowson M 25 Alarcon M, Cantor RM, Liu J, Gilliam TC, Geschwind DH. et al. A case-control family history study of autism. J Child Psychol Evidence for a language quantitative trait locus on chromosome Psychiatry 1994; 35: 877–900. 7q in multiplex autism families. Am J Hum Genet 2002; 70: 60–71. 4 Lamb JA, Parr JR, Bailey AJ, Monaco AP. Autism: in search of 26 Bradford Y, Haines J, Hutcheson H, Gardiner M, Braun T, Sheffield susceptibility genes. Neuromolecular Med 2002; 2: 11–28. V et al. Incorporating language phenotypes strengthens evidence 5 Auranen M, Vanhala R, Varilo T, Ayers K, Kempas E, Ylisaukko-Oja T of linkage to autism. Am J Med Genet 2001; 105: 539–547. et al. A genomewide screen for autism-spectrum disorders: 27 Sutcliffe JS, Delahanty RJ, Prasad HC, McCauley JL, Han Q, Jiang L evidence for a major susceptibility locus on chromosome 3q25– et al. Allelic heterogeneity at the serotonin transporter locus 27. Am J Hum Genet 2002; 71: 777–790. (SLC6A4) confers susceptibility to autism and rigid-compulsive 6 Buxbaum JD, Silverman JM, Smith CJ, Kilifarski M, Reichert J, behaviors. Am J Hum Genet 2005; 77: 265–279. Hollander E et al. Evidence for a susceptibility gene for autism on 28 Shao Y, Cuccaro ML, Hauser ER, Raiford KL, Menold MM, chromosome 2 and for genetic heterogeneity. Am J Hum Genet Wolpert CM et al. Fine mapping of autistic disorder to 2001; 68: 1514–1520. chromosome 15q11–q13 by use of phenotypic subtypes. Am 7 Liu J, Nyholt DR, Magnussen P, Parano E, Pavone P, Geschwind D J Hum Genet 2003; 72: 539–548. et al. A genomewide screen for autism susceptibility loci. 29 Coon H, Matsunami N, Stevens J, Miller J, Pingree C, Camp NJ Am J Hum Genet 2001; 69: 327–340. et al. Evidence for linkage on chromosome 3q25–27 in a large 8 McCauley JL, Li C, Jiang L, Olson LM, Crockett G, Gainer K et al. autism extended pedigree. Hum Hered 2006; 60: 220–226. Genome-wide and Ordered-Subset linkage analyses provide sup- 30 Lord C, Risi S, Lambrecht L, Cook Jr EH, Leventhal BL, DiLavore port for autism loci on 17q and 19p with evidence of phenotypic PC et al. The autism diagnostic observation schedule-generic: a and interlocus genetic correlates. BMC Med Genet 2005; 6:1. standard measure of social and communication deficits associated 9 Risch N, Spiker D, Lotspeich L, Nouri N, Hinds D, Hallmayer J with the spectrum of autism. J Autism Dev Disord 2000; 30: et al. A genomic screen of autism: evidence for a multilocus 205–223. etiology. Am J Hum Genet 1999; 65: 493–507. 31 Lord C, Rutter M, Le Couteur A. Autism Diagnostic Interview- 10 Barrett S, Beck JC, Bernier R, Bisson E, Braun TA, Casavant TL Revised: a revised version of a diagnostic interview for caregivers et al. An autosomal genomic screen for autism. Collaborative of individuals with possible pervasive developmental disorders. linkage study of autism. Am J Med Genet 1999; 88: 609–615. J Autism Dev Disord 1994; 24: 659–685. 11 Cantor RM, Kono N, Duvall JA, Alvarez-Retuerto A, Stone JL, 32 Wechsler D. Wechsler Adult Intelligence Scale—Third Edition. Alarcon M et al. Replication of autism linkage: fine-mapping peak The Psychological Corporation: San Antonio, TX, 1997. at 17q21. Am J Hum Genet 2005; 76: 1050–1056. 33 Wechsler D. Manual for the Wechsler Intelligence Scale for 12 International Molecular Genetic Study of Autism Consortium Children—Third Edition. The Psychological Corporation: San (IMGSAC). A genomewide screen for autism: strong evidence for Antonio, TX, 1991. linkage to chromosomes 2q, 7q, and 16p. Am J Hum Genet 2001; 34 Elliott CD. Differential Ability Scales. The Psychological Corpora- 69: 570–581. tion: San Antonio, TX, 1990. 13 Philippe A, Martinez M, Guilloud-Bataille M, Gillberg C, 35 Jorde LB. Inbreeding in the Utah Mormons: an evaluation of Rastam M, Sponheim E et al. Genome-wide scan for autism estimates based on pedigrees, isonymy, and migration matrices. susceptibility genes. Paris Autism Research International Sibpair Ann Hum Genet 1989; 53(Part 4): 339–355. Study. Hum Mol Genet 1999; 8: 805–812. 36 Jorde LB. Consanguinity and prereproductive mortality in the 14 Shao Y, Wolpert CM, Raiford KL, Menold MM, Donnelly SL, Utah Mormon population. Hum Hered 2001; 52: 61–65. Ravan SA et al. Genomic screen and follow-up analysis for autistic 37 Matsuzaki H, Loi H, Dong S, Tsai YY, Fang J, Law J et al. Parallel disorder. Am J Med Genet 2002; 114: 99–105. genotyping of over 10 000 SNPs using a one-primer assay on a 15 Lauritsen MB, Als TD, Dahl HA, Flint TJ, Wang AG, Vang M et al. high-density oligonucleotide array. Genome Res 2004; 14: A genome-wide search for alleles and haplotypes associated with 414–425. autism and related pervasive developmental disorders on the 38 Mukhopadhyay N, Almasy L, Schroeder M, Mulvihill WP, Weeks DE. Faroe Islands. Mol Psychiatry 2006; 11: 37–46. Mega2: data-handling for facilitating genetic linkage and associa- 16 Schellenberg GD, Dawson G, Sung YJ, Estes A, Munson J, tion analyses. Bioinformatics 2005; 21: 2556–2557. Rosenthal E et al. Evidence for multiple loci from a genome scan 39 Thomas A. GMCheck: Bayesian error checking for pedigree of autism kindreds. Mol Psychiatry 2006; 11: 1049–1060. genotypes and phenotypes. Bioinformatics 2005; 21: 3187–3188. 17 Szatmari P, Paterson AD, Zwaigenbaum L, Roberts W, Brian J, Liu XQ 40 Thomas A, Gutin A, Abkevich V, Bansai A. Multipoint linkage et al. Mapping autism risk loci using genetic linkage and analysis by blocked (Gibbs) sampling. Stats Comp 2000; 10: chromosomal rearrangements. Nat Genet 2007; 39: 319–328. 259–269.

Molecular Psychiatry High-density linkage in large autism pedigree K Allen-Brady et al 600 41 Abkevich V, Camp NJ, Hensel CH, Neff CD, Russell DL, families with autism spectrum disorders originating from a Hughes DC et al. Predisposition locus for major depression subisolate of Finland. Mol Psychiatry 2003; 8: 879–884. at chromosome 12q22–12q23.2. Am J Hum Genet 2003; 73: 54 Ylisaukko-oja T, Alarcon M, Cantor RM, Auranen M, Vanhala R, 1271–1281. Kempas E et al. Search for autism loci by combined analysis of 42 Camp NJ, Farnham JM, Cannon Albright LA. Genomic search for Autism Genetic Resource Exchange and Finnish families. Ann prostate cancer predisposition loci in Utah pedigrees. Prostate Neurol 2006; 59: 145–155. 2005; 65: 365–374. 55 Limprasert P, Zhong N, Dobkin C, Brown WT. Polymorphism of 43 Camp NJ, Lowry MR, Richards RL, Plenk AM, Carter C, Hensel CH FXR1 showing lack of association with autism. Am J Med Genet et al. Genome-wide linkage analyses of extended Utah pedigrees 1997; 74: 453–454. identifies loci that influence recurrent, early-onset major depres- 56 Ylisaukko-oja T, Rehnstrom K, Auranen M, Vanhala R, Alen R, sion and anxiety disorders. Am J Med Genet B Neuropsychiatr Kempas E et al. Analysis of four neuroligin genes as candidates for Genet 2005; 135: 85–93. autism. Eur J Hum Genet 2005; 13: 1285–1292. 44 Camp NJ, Neuhausen SL, Tiobech J, Polloi A, Coon H, 57 Wolpert CM, Donnelly SL, Cuccaro ML, Hedges DJ, Poole CP, Myles-Worsley M. Genomewide multipoint linkage analysis of Wright HH et al. De novo partial duplication of chromosome 7p in seven extended Palauan pedigrees with schizophrenia, by a a male with autistic disorder. Am J Med Genet 2001; 105: 222–225. Markov-chain Monte Carlo method. Am J Hum Genet 2001; 69: 58 Rodier PM, Ingram JL, Tisdale B, Croog VJ. Linking etiologies in 1278–1289. humans and animal models: studies of autism. Reprod Toxicol 45 Thomas A, Camp NJ. Maximum likelihood estimates of allele 1997; 11: 417–422. frequencies and error rates from samples of related individuals by 59 Ingram JL, Stodgell CJ, Hyman SL, Figlewicz DA, Weitkamp LR, gene counting. Bioinformatics 2006; 22: 771–772. Rodier PM. Discovery of allelic variants of HOXA1 and HOXB1: 46 Schaid DJ, Guenther JC, Christensen GB, Hebbring S, Rosenow C, genetic susceptibility to autism spectrum disorders. Teratology Hilker CA et al. Comparison of microsatellites versus 2000; 62: 393–405. single-nucleotide polymorphisms in a genome linkage screen for 60 Conciatori M, Stodgell CJ, Hyman SL, O’Bara M, Militerni R, prostate cancer-susceptibility Loci. Am J Hum Genet 2004; 75: Bravaccio C et al. Association between the HOXA1 A218G 948–965. polymorphism and increased head circumference in patients with 47 Amos CI, Chen WV, Lee A, Li W, Kern M, Lundsten R et al. autism. Biol Psychiatry 2004; 55: 413–419. High-density SNP analysis of 642 Caucasian families with 61 Ronald A, Happe F, Bolton P, Butcher LM, Price TS, Wheelwright rheumatoid arthritis identifies two new linkage regions on 11p12 S et al. Genetic heterogeneity between the three components of the and 2q33. Genes Immun 2006; 7: 277–286. autism spectrum: a twin study. J Am Acad Child Adolesc 48 Horne BD, Camp NJ. Principal component analysis for selection of Psychiatry 2006; 45: 691–699. optimal SNP-sets that capture intragenic genetic variation. Genet 62 Lamb JA, Barnby G, Bonora E, Sykes N, Bacchelli E, Blasi F et al. Epidemiol 2004; 26: 11–21. Analysis of IMGSAC autism susceptibility loci: evidence for sex 49 Lander E, Kruglyak L. Genetic dissection of complex traits: limited and parent of origin specific effects. J Med Genet 2005; 42: guidelines for interpreting and reporting linkage results. Nat 132–137. Genet 1995; 11: 241–247. 63 Boyles AL, Scott WK, Martin ER, Schmidt S, Li YJ, Ashley-Koch A 50 Ray A, Weeks DE. No convincing evidence of linkage for restless et al. Linkage disequilibrium inflates type I error rates in legs syndrome on chromosome 9p. Am J Hum Genet 2005; 76: multipoint linkage analysis when parental genotypes are missing. 705–707; author reply 7–10. Hum Hered 2005; 59: 220–227. 51 Terwilliger JD, Speer M, Ott J. Chromosome-based method for 64 Sellick GS, Webb EL, Allinson R, Matutes E, Dyer MJ, Jonsson V rapid computer simulation in human genetic linkage analysis. et al. A high-density SNP genomewide linkage scan for chronic Genet Epidemiol 1993; 10: 217–224. lymphocytic leukemia-susceptibility loci. Am J Hum Genet 2005; 52 Martineau J, Herault J, Petit E, Guerin P, Hameury L, Perrot A et al. 77: 420–429. Catecholaminergic metabolism and autism. Dev Med Child Neurol 65 John S, Shephard N, Liu G, Zeggini E, Cao M, Chen W et al. Whole- 1994; 36: 688–697. genome scan, in a complex disease, using 11 245 single-nucleotide 53 Auranen M, Varilo T, Alen R, Vanhala R, Ayers K, Kempas E et al. polymorphisms: comparison with microsatellites. Am J Hum Evidence for allelic association on chromosome 3q25–27 in Genet 2004; 75: 54–64.

Molecular Psychiatry