Genome-wide association studies (GWAS) in schizophrenia have focused on additive allelic effects to identify disease risk loci. In order to examine potential recessive effects, we applied a novel approach to identify regions of excess homozygosity in an ethnically homogenous cohort: 904 schizophrenia cases and 1640 controls drawn from the Ashkenazi Jewish (AJ) population. Genome-wide examination of runs of homozygosity identified an excess in cases localized to the major histocompatibility complex (MHC). To refine this signal, we used the recently developed GERMLINE algorithm to identify chromosomal segments shared identical-by-descent (IBD) and compared homozygosity at such segments in cases and controls. We found a significant excess of homozygosity in schizophrenia cases compared with controls in the MHC (P-value = 0.003). An independent replication cohort of 548 schizophrenia cases from Japan and 542 matched healthy controls demonstrated similar effects. The strongest case–control recessive effects (P = 8.81 × 10−8) were localized to a 53-kb region near HLA-A, in a segment encompassing three poorly annotated genes, TRIM10, TRIM15 and TRIM40. At the same time, an adjacent segment in the Class I MHC demonstrated clear additive effects on schizophrenia risk, demonstrating the complexity of association in the MHC and the ability of our IBD approach to refine localization of broad signals derived from conventional GWAS. In sum, homozygosity in the classical MHC region appears to convey significant risk for schizophrenia, consistent with the ecological literature suggesting that homozygosity at the MHC locus may be associated with vulnerability to disease.

INTRODUCTION

Schizophrenia (SCZ) is a severe and chronic psychiatric disorder that affects ∼1% of the global population. It is highly heritable (0.7–0.8) (1), but its complex pattern of inheritance has hindered efforts to identify susceptibility genes. With the advancement of high-throughput single-nucleotide genotyping technologies, genome-wide association studies (GWAS) have discovered the first robust and replicable risk loci for SCZ; however, these SNPs account for only a tiny proportion of the total genetic risk (2). Moreover, GWAS generally examine only common SNPs with additive allelic effects on disease susceptibility, but such effects contribute only about one-quarter of the variance in liability to SCZ (3). Therefore, complementary approaches are required to identify sources of the ‘missing heritability’ of this phenotype (4,5).

An alternative approach for investigating genetic association is homozygosity mapping (6). Although homozygosity mapping was originally developed to identify rare recessive effects in consanguineous families (7), long chromosomal segments of consecutive homozygous alleles can be found throughout the genome even in outbred individuals (8). These ‘runs of homozygosity’ (ROHs) represent a form of genetic variation that can be examined for associations with phenotypic variation, which are not additive and thus are poorly assessed by standard GWAS analysis. Such ROHs may harbor rare deleterious variants, and even in outbred populations, an excess of ROHs may produce negative phenotypic effects owing to a process similar to inbreeding depression (9,10). Recent evidence has shown that a genome-wide excess of ROHs may be associated with increased risk of SCZ (11,12), as well as Alzheimer's disease (13). Specific regions of excess homozygosity have further been implicated in risk for Alzheimer's disease (13), rheumatoid arthritis (14) and autism (15).

Recent studies of ROHs in SCZ and other complex disorders have been marked by several limitations. First, ROHs are sparsely distributed among individuals from outbred populations, imposing substantial limitations on power (12). Second, the relative presence of ROHs can be impacted by subtle differences in ancestry that are difficult to control for in heterogeneous populations, such as European-Americans (16,17). Third, commonly used methods for the identification of ROHs typically do not distinguish between different alleles that may be homozygous at a given locus (11). Finally, the presence of ROHs may be a function of common haplotypes and/or lack of recombination at a given chromosomal locus, rather than having any association with disease; studies to-date have not directly compared ROH prevalence against baseline levels of chromosomal sharing. Recently developed methods utilizing hidden Markov models to determine identity-by-descent (IBD) of chromosomal segments permit a more precise specification of true autozygosity (1820).

The present study was designed to overcome the aforementioned limitations by examining ROHs in an SCZ case–control cohort drawn from a homogeneous Ashkenazi Jewish (AJ) founder population (21). Further, we utilized two complementary methods to examine ROHs, using both segment length and inferred IBD status to determine ROH thresholds. In our cohort of 904 AJ SCZ cases and 1640 AJ controls, genotyped for ∼1 million SNPs (Illumina Omni1-Quad platform), we compared the presence of ROHs between cases and controls using our previously published whole-genome homozygosity association (WGHA) method (11). We then further refined this signal using an IBD approach: we compared the presence of homozygous or heterozygous tracts of IBD both genome-wide as well as on a locus-by-locus basis. Finally, we sought to replicate our findings in an independent SCZ case–control cohort from a different ethnically homogeneous (though not founder) population from Japan.

RESULTS

Principal component analysis (PCA) using ancestry informative markers (AIMs) that we have previously identified specifically for the AJ population (21) showed our AJ data set of 904 SCZ cases and 1640 controls to form a tight cluster (22); similar results are obtained with standard genome-wide PCA. After this QC, we performed WGHA in the AJ cohort. Specifically, we focused on large (>1.5 Mb in length with minimum 300 markers) ROHs for two reasons: (1) It has been shown that these are more likely to harbor rare, disease-associated variants (10,12) and (2) a comprehensive examination of ROHs across multiple worldwide populations has demonstrated that ∼1.5 Mb represents a reliable threshold for identifying ROHs that are due to true autozygosity (homozygosity by descent) (16). Comparing the tally of ROHs in cases versus controls at each locus, a genome-wide significant signal for excess ROHs within cases was observed at the MHC locus (P-value = 1.076 × 10−5, FDR = 0.0022) (Fig. 1). However, the observed peak was quite broad, extending ∼3 Mb (all P < 0.01 at hg18 coordinates chr6: 30053249–32848389).

Figure 1.

Manhattan plot for ROH analysis. Genome-wide significance threshold was set to P < 2.12 × 10−5 owing to non-independence across SNPs (mean ∼500 SNPs/ROH). The genome-wide significant signal was localized to the chr6 MHC region.

Figure 1.

Manhattan plot for ROH analysis. Genome-wide significance threshold was set to P < 2.12 × 10−5 owing to non-independence across SNPs (mean ∼500 SNPs/ROH). The genome-wide significant signal was localized to the chr6 MHC region.

We next sought to refine localization of this signal using the GERMLINE algorithm (18). We also sought to determine whether these excess ROHs reflected specific haplotypes that are shared IBD between significantly larger numbers of case haplotype carriers compared with control haplotype carriers. We converted pairwise IBD segments into multi-sample haplotype clusters by using the DASH clustering algorithm (20). In our AJ data set, DASH analysis identified ∼1 million haplotype clusters genome wide. Each subject was then classified as a heterozygous carrier, homozygous carrier or non-carrier of each IBD segment, and the total number of heterozygous and homozygous IBD segments detected were counted for each subject.

In order to compare rates of homozygous IBD sharing in cases and controls, it was first necessary to control for baseline rates of IBD sharing, which can vary as a function of subtle sampling differences in ancestry (18). Our hypothesis therefore tests the rate of homozygous IBD sharing, controlling for overall rates of heterozygous IBD sharing using ANCOVA (Tables 1 and 2). Across the entire genome, we observed that the rate of carriage of homozygous IBD segments (controlling for the rate of heterozygous IBD sharing) was not significantly increased in SCZ cases compared with controls (Table 1) For the extended MHC region (Chr6: 25–35 Mb), however, we observed a significantly increased rate of homozygous IBD segments in cases relative to controls (P-value = 0.003; Table 1). Both MHC Class I and II regions show an excess of homozygous IBD segments in SCZ cases (Table 2), although the latter comparison only trends toward statistical significance owing to the lower overall rate of IBD sharing in that region. Effect sizes were small, but comparable, across each of the MHC segments (Cohen's d = 0.123 for the entire MHC, 0.118 for the Class I MHC and 0.082 for the Class II MHC).

Table 1.

Comparison of homozygous and heterozygous IBD segments across the whole genome and extended MHC region (chr6: 25–35 Mb) in AJ patients and controls

 Whole genome
 
MHC locus chr6: 25–35 MB 
Cases Controls Cases Controls 
Homozygous IBD segments 
 Mean per subject (SD) 30.97 (24.19) 29.93 (21.13) 10.23 (17.16) 8.35 (13.08) 
Heterozygous IBD segments 
 Mean per subject (SD) 7157.9 (913.0) 7,258.6 (715.8) 123.2 (43.4) 125.14 (42.2) 
ANCOVA (Hom controlling Het) F1,2541 = 2.56 P = 0.110 F1,2541 = 8.72 P = 0.003 
 Whole genome
 
MHC locus chr6: 25–35 MB 
Cases Controls Cases Controls 
Homozygous IBD segments 
 Mean per subject (SD) 30.97 (24.19) 29.93 (21.13) 10.23 (17.16) 8.35 (13.08) 
Heterozygous IBD segments 
 Mean per subject (SD) 7157.9 (913.0) 7,258.6 (715.8) 123.2 (43.4) 125.14 (42.2) 
ANCOVA (Hom controlling Het) F1,2541 = 2.56 P = 0.110 F1,2541 = 8.72 P = 0.003 
Table 2.

Comparison of homozygous and heterozygous IBD segments for MHC Class I and II regions in AJ patients and controls

 MHC Class I region
 
MHC Class II region
 
Cases Controls Cases Controls 
Homozygous IBD segments 
 Mean per subject (SD) 4.36 (9.59) 3.35 (7.38) 0.60 (2.50) 0.42 (1.86) 
Heterozygous IBD segments 
 Mean per subject (SD) 57.12 (24.33) 57.67 (23.16) 17.22 (10.07) 17.73 (9.94) 
ANCOVA (Hom controlling Het) F1,2541 = 8.498 P = 0.004 F1,2541 = 3.388 P = 0.066 
 MHC Class I region
 
MHC Class II region
 
Cases Controls Cases Controls 
Homozygous IBD segments 
 Mean per subject (SD) 4.36 (9.59) 3.35 (7.38) 0.60 (2.50) 0.42 (1.86) 
Heterozygous IBD segments 
 Mean per subject (SD) 57.12 (24.33) 57.67 (23.16) 17.22 (10.07) 17.73 (9.94) 
ANCOVA (Hom controlling Het) F1,2541 = 8.498 P = 0.004 F1,2541 = 3.388 P = 0.066 

As a replication cohort, we used data from a Japanese GWAS of 548 patients with SCZ and 542 matched healthy controls with 265,558 markers after QC (23). All subjects were selected from the Tokai area of the Honshu Island of Japan. Previous studies have shown that this is an extremely homogeneous population, with slightly elevated rates of homozygosity compared with European cohorts (16). As presented in Table 3, we again observed higher homozygous IBD sharing in cases compared with controls across the extended MHC region, with comparable effect sizes as observed in the Ashkenazi cohort. It is important to note that the overall (genome wide) rate of homozygous IBD sharing in the outbred (though homogeneous) Japanese cohort was 30-fold lower than that in the Ashkenazi cohort, greatly reducing statistical power. However, meta-analysis revealed a robustly significant effect across the two cohorts (P = 0.0012).

Table 3.

Comparison of homozygous IBD segments across the extended MHC region (chr6: 25–35 Mb) in AJ and Japanese cohorts

Homozygous IBD segments at MHC locus chr6: 25–35 Mb
 
 Cases Controls Effect sizea P-value 
Ashkenazi Jewish Cohort 
 Mean per subject (SD) 10.23 (17.16) 8.35 (13.08) 0.123 0.003b 
Japanese Cohort 
 Mean per subject (SD) 0.34 (1.66) 0.23 (1.30) 0.074 0.196b 
   Meta-analysis 0.0012c 
Homozygous IBD segments at MHC locus chr6: 25–35 Mb
 
 Cases Controls Effect sizea P-value 
Ashkenazi Jewish Cohort 
 Mean per subject (SD) 10.23 (17.16) 8.35 (13.08) 0.123 0.003b 
Japanese Cohort 
 Mean per subject (SD) 0.34 (1.66) 0.23 (1.30) 0.074 0.196b 
   Meta-analysis 0.0012c 

aCohen's d comparison between means of cases and controls.

bP-value determined by ANCOVA, controlling for the rate of heterozygous segments.

cMeta-analytic P-value calculated using Stouffer's z trend procedure.

We next focused on individual homozygous IBD segments within the AJ SCZ cohort to pinpoint the source of the MHC signal. We performed association tests for IBD segments within the MHC, comparing additive, dominant and recessive models. Only two segments met the MHC-wide significance threshold (P < 1.09 × 10−4) for any model. As displayed in Table 4, one segment located at chr6: 30206368–30259349 (hg18) was strongly significant under the recessive model (chi-square = 28.62, df = 1, P = 8.81 × 10−8). Compared with the recessive model, the chi-square value for this same segment under the additive model was only half as large with P-value more than 3 orders of magnitude larger (chi-square = 14.47, df = 1, P = 1.42 × 10−4). This result is especially striking given that the additive model has enhanced power owing to measuring chromosomes (2n) rather than individuals (n). As displayed in Figure 2, this segment localized to the Class I MHC and encompassed three poorly annotated genes, TRIM10, TRIM15 and TRIM40. In contrast, a neighboring segment (chr6: 30148063–30206302) in the Class I MHC demonstrated evidence of MHC-wide significant additive effects (chi-square = 20.19, df = 1, P = 7.03 × 10−6), without evidence of strong recessive effects (Table 5); this segment contains two genes (RNF39 and TRIM31). No other IBD segments met the MHC-wide significance threshold under any model. Supplementary Material, Table S1, displays nine additional segments that approached significance (nominal P < 0.001 for any model); of these, seven segments had the strongest effects under the additive model, whereas two demonstrated strongest effects under the recessive model. These segments spanned both Class I and Class II MHC loci.

Table 4.

Association test for IBD region localized to chr6: 30206368–30259349

Model Allele count cases (carrier/non) Allele count controls (carrier/non) Chi-squared statistic Degrees of freedom P-value 
Additive model 352/1456 502/2778 14.47 1.42 × 10−4 
Dominant model 306/598 478/1162 6.05 0.01394 
Recessive model 46/858 24/1616 28.62 8.81 × 10−8 
Model Allele count cases (carrier/non) Allele count controls (carrier/non) Chi-squared statistic Degrees of freedom P-value 
Additive model 352/1456 502/2778 14.47 1.42 × 10−4 
Dominant model 306/598 478/1162 6.05 0.01394 
Recessive model 46/858 24/1616 28.62 8.81 × 10−8 
Table 5.

Association test for IBD region localized to chr6: 30148063–30206302

Model Allele count cases (carrier/non) Allele count controls (carrier/non) Chi-squared statistic Degrees of freedom P-value 
Additive model 273/1535 355/2925 19.7 9.06 × 10−6 
Dominant model 249/655 340/1300 15.2 9.66 × 10−5 
Recessive model 24/880 15/1625 11.69 6.28 × 10−4 
Model Allele count cases (carrier/non) Allele count controls (carrier/non) Chi-squared statistic Degrees of freedom P-value 
Additive model 273/1535 355/2925 19.7 9.06 × 10−6 
Dominant model 249/655 340/1300 15.2 9.66 × 10−5 
Recessive model 24/880 15/1625 11.69 6.28 × 10−4 
Figure 2.

The most significant homozygous IBD segment under the recessive model (shown as red bar) that was associated with SCZ. The top signal was localized to TRIM10, TRIM15 and TRIM40. The green bar represents the IBD segment with the strongest P-value under the additive model, localized to RNF39 and TRIM31.

Figure 2.

The most significant homozygous IBD segment under the recessive model (shown as red bar) that was associated with SCZ. The top signal was localized to TRIM10, TRIM15 and TRIM40. The green bar represents the IBD segment with the strongest P-value under the additive model, localized to RNF39 and TRIM31.

DISCUSSION

The major histocompatibility complex (MHC) has emerged as a region of major interest in SCZ genetics. The first evidence supporting the MHC as a potential SCZ susceptibility locus dates back to 1974 (24). More recently, large-scale GWAS studies have converged to demonstrate that the MHC contains the strongest association signal (2,2528). Importantly, prior GWAS studies have been unable to precisely localize the source of this signal, owing to the extensive long-range linkage disequilibrium throughout the MHC; different studies have identified top SNPs ranging across a nearly 10-Mb extent (coordinates ranging from 25 to 35 Mb on Chromosome 6). Additionally, prior association studies of the MHC have exclusively examined an additive model of risk, in which each additional allele at a given locus contributes to risk in a linear manner.

The present study has extended prior research on the MHC in SCZ in two ways. First, we identified a global increase in MHC homozygosity in SCZ cases compared with controls. Second, utilizing an IBD approach in our founder population, we were able to further localize this effect to at least one specific segment in the Class I MHC; the strongest case–control recessive effects were observed near the HLA-A gene, in a segment encompassing three poorly annotated genes, TRIM10, TRIM15 and TRIM40. A run of homozygosity in this region, while relatively rare (<2% frequency in AJ controls), conveyed a 3-fold increase in the risk for SCZ. At the same time, we observed that these recessive effects coexist side-by-side with other association signals within the MHC that are more consistent with an additive effect. In particular, a segment encompassing RNF39 and TRIM31 demonstrated significant additive effects in our Ashkenazi cohort. While these genes are also poorly annotated, it is noteworthy that a prior GWAS has detected a strong additive association with the levels of the MHC component molecule beta-2 microglobulin to an SNP in TRIM31 (29).

Prior GWAS implicating the MHC in SCZ have been interpreted in the context of growing evidence that immune system dysfunction might play an important role in the etiology of SCZ. For example, in epidemiologic studies, several MHC-linked autoimmune/inflammatory disorders are found to have significantly increased prevalence in patients with SCZ (and their relatives) (3032); moreover, exposure to prenatal infections has repeatedly been found to enhance risk of developing SCZ (3336). Additionally, numerous studies have detected elevated levels of inflammatory markers (e.g. IL-2 and TNF-α) associated with immune system activation in patients with SCZ (3739). It is important to emphasize that the present report, as with all genetic association studies, is not able to determine the mechanism by which MHC homozygosity affects risk; further research is needed in this area.

To the extent that immune mechanisms might be mediating the observed increase in homozygosity, however, it may be relevant to consider the heterozygote advantage hypothesis of the MHC. This hypothesis proposes that individuals heterozygous at MHC loci are able to respond to a greater range of pathogen peptides than homozygous individuals, and consequently, are more likely to have higher relative fitness (4042). However, other possible explanations cannot be ruled out: (1) the specific rare alleles identified in our study may increase susceptibility or (2) excess homozygosity may mark individuals with increased risk owing to maternal–fetal HLA matching (43). It should be noted that a negative result has been reported (44) by a recent study using a large, well-characterized trio cohort to determine whether non-random mating, parent-of-origin effects for HLA alleles and maternal–fetal genotype incompatibility in the HLA increased risk for SCZ. Because our samples were drawn from a case–control cohort, further studies would be required to test this hypothesis more directly.

At the same time, many genes within the MHC are not primarily related to immune function and point to other possible neurodevelopmental mechanisms including synaptic plasticity, neurotransmitter signaling and brain myelination (45). For example, the class I HLA region includes two non-immune-related candidate genes for SCZ: MOG, which encodes myelin oligodendrocyte glycoprotein, a critical constituent of the myelin sheath which may be aberrant in SCZ (46), and GABBR1, which encodes a subunit of the metabotropic GABA(B) receptor, which is also implicated in SCZ pathophysiology (47). Moreover, HLA molecules themselves have recently been demonstrated to be directly expressed in the brain at varying levels during early development, even in the absence of infection or insult (48). In a recent study, the carriers of an SCZ-associated risk allele in the Class I MHC demonstrated larger ventricular volume (49), consistent with prior evidence of enlarged ventricles in Class I-deficient mice (50). It should be noted that the genes identified in our top recessive segment (TRIM10, TRIM15 and TRIM40) are only weakly expressed in mouse brain and are variably expressed in adult human brain according to the Allen Brain Atlas. Therefore, the limited knowledge on the biological function of TRIM10, TRIM15 and TRIM40 does not allow extensive speculation on the role of these genes in the etiology of the disease.

SUBJECTS AND METHOD

Participants

A total of 1156 SCZ cases from ethnically homogenous population (Ashkenazi Jews) have been collected and ascertained at the Hebrew University, Israel (Hebrew University Genetic Resource, HUGR, http://hugr.huji.ac.il). Cases were recruited from hospitalized inpatients at seven medical centers in Israel. All diagnoses were assigned after direct interview using the structured clinical interview, a questionnaire with inclusion and exclusion criteria, and cross-references to medical records. The inclusion criteria specified that subjects had to be diagnosed with SCZ or schizoaffective disorder by the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV), that all four grandparents of each subject were reported by the subject to be of AJ ethnic origin, and that each subject or the subject's legal representative has signed the informed consent form. The exclusion criteria eliminated subjects diagnosed with at least one of the following disorders: psychotic disorder owing to a general medical condition, substance-induced psychotic disorder or any Cluster A (schizotypal, schizoid or paranoid) personality disorder. Control samples (n = 2279) from healthy Ashkenazi individuals were collected from volunteers at the Israeli Blood Bank; these subjects were not psychiatrically screened but reported no chronic disease and were taking no medication at the time of blood draw. Corresponding institutional review boards and the National Genetic Committee of the Israeli Ministry of Health approved the studies. All samples were fully anonymized immediately after collection.

As a replication cohort, we used data from a Japanese SCZ GWAS of 575 patients with SCZ and 564 matched healthy controls as previously described (23). All subjects were unrelated, living in the Tokai area of the mainland of Japan, and self-identified as Japanese. Patients were included if they (1) met DSM-IV criteria for SCZ and (2) were physically healthy. Patients were excluded if they had a history of substance abuse, neurodevelopmental disorders, epilepsy or known mental retardation. Consensus diagnoses were made by at least two experienced psychiatrists according to DSM-IV criteria on the basis of unstructured interviews with patients, their families and review of medical records. Controls were selected from the general population with no history of mental disorders based upon self-report. After description of the study, written informed consent was obtained from each subject. This study was approved by the ethics committees of each participating university.

Genotyping and quality control

Ashkenazi Jewish samples were genotyped on 1 million genome-wide SNPs using Illumina HumanOmni1-Quad arrays according to manufacturers' specifications. SNP data for the Japanese SCZ cohort were obtained using the Affymetrix Human SNP array 5.0. To obtain high-quality data for further analysis, we performed QC steps for both data sets as follows: SNPs were filtered on call rate <98%, minor allele frequency <0.02 and Hardy–Weinberg exact test P < 0.000001 in controls. Samples were filtered based on genotype quality control filtration (sample call rate <97%, gender mismatch) and examined for cryptic identity and first- or second-degree relatedness using pairwise IBD estimation (PI_HAT) in PLINK (51) with 128,403 LD pruned (r2 > 0.2) genome-wide SNPs. Samples were excluded based on PI_Hat >0.125; the individual with lower call rate from each control/control or case/case pair was excluded, and controls were excluded from case–control pairs. After QC, there were 762,372 and 265,558 markers obtained for the AJ data set and Japan data set, respectively.

Population stratification

To obtain a homogenous population, the AJ samples after QC were further examined for underlying population stratification using PCA with AIMs specific for the AJ population (21). Samples with PCA results suggestive of one or more non-AJ grandparents were identified as outliers based on first principal component score of >0.01 and were excluded from further analysis (n = 607). We then compared the remaining AJ samples with HapMap CEU, JPT and CHB and YRI population. Following sample quality control filters and PCA-based population homogeneity, a total of 904 AJ SCZ cases and 1640 AJ controls as well as 548 Japanese SCZ cases and 542 Japanese controls were included in our IBD analysis. All PCA plots were obtained using the SVS7 (Golden Helix, Inc.) software package.

Runs of homozygosity

We performed WGHA using the SVS7 software package (GoldenHelix, Inc.) following methods described previously (11). Briefly, SNP data from each chromosome of each subject were interrogated for ROHs, which are long series of consecutive SNPs that are homozygous. Based on prior literature (10,12), we were interested in long, rare ROHs, so minimum run length was set to 1500 kb (with minimum 300 SNPs and no gap >100 kb to exclude centromeres and other anomalies). To permit some tolerance for genotyping error, up to one heterozygote call and up to five missing genotypes were permitted within a called ROH. After the identification of ROHs, each subject's SNP data were then converted to binary calls (0 or 1) at each position indicating whether that SNP is a member of a ROH for that individual. Next, at each position, data from all subjects were examined to determine how many cases and controls share a ROH call at a given position. There were 708,433 SNP positions, which were part of an ROH for >1 individual in the cohort; frequency of ROH in cases versus controls was compared at each such position using chi-square tests. Given that at least 300 consecutive SNPs were to comprise a ROH call, a threshold of P < 0.05/(708433/300) or P < 2.12 × 10−5 was required for genome-wide significance.

Identity-by-descent and homozygosity detection

As a complementary approach to identifying homozygosity, we utilized the GERMLINE algorithm (18), which uses a dynamically adjusted sliding window to detect, across all pairs of samples, genomic segments shared IBD from a recent ancestor. Genotype data were first phased using Beagle (version 3.3.2) (52,53) with default parameters. The phased genotype data were interrogated for pairwise shared IBD segments using GERMLINE (18). We ran GERMLINE with parameters tuned to identify IBD segments of 3 cM or greater, with a window size of 128 sites and 3 mismatching homozygous sites allowed. We next executed DASH (20) on the IBD segments to identify the specific cases and controls sharing each particular IBD segment. Using in-house perl scripts, we categorized each segment in each subject as homozygous or heterozygous and compared rates of homozygous IBD segments in cases versus controls, controlling for rates of heterozygous sharing, using ANCOVA (implemented in SPSS 22).

After conducting global tests of homozygous versus heterozygous IBD segments, we sought to localize the source of signal within the MHC by performing association tests in PLINK. For each of 457 IBD segments tested in the MHC region, we compared allelic, dominant and recessive models. We sought to identify segments that attained significance below a Bonferroni-corrected threshold (0.05/457 = 1.09 × 10−4) for any of these models. For each significant segment, we further sought to determine the best fitting model.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported in part by grants from the NIH to S.M. (K99 MH101255), T.L. (RC2 MH089964; R01 MH084098) and A.M. (P50 MH080173). S.M. has received compensation for consultation from Genomind, Inc.

REFERENCES

1
Sullivan
P.F.
Kendler
K.S.
Neale
M.C.
Schizophrenia as a complex trait: evidence from a meta-analysis of twin studies
Arch. Gen. Psychiatry
 
2003
60
1187
1192
2
Ripke
S.
Sanders
A.R.
Kendler
K.S.
Levinson
D.F.
Sklar
P.
Holmans
P.A.
Lin
D.Y.
Duan
J.
Ophoff
R.A.
Andreassen
O.A.
et al.  
Genome-wide association study identifies five new schizophrenia loci
Nat. Genet.
 
2011
43
969
976
3
Lee
S.H.
Decandia
T.R.
Ripke
S.
Yang
J.
Sullivan
P.F.
Goddard
M.E.
Keller
M.C.
Visscher
P.M.
Wray
N.R.
Estimating the proportion of variation in susceptibility to schizophrenia captured by common SNPs
Nat. Genet.
 
2012
44
831
4
Mowry
B.J.
Gratten
J.
The emerging spectrum of allelic variation in schizophrenia: current evidence and strategies for the identification and functional characterization of common and rare variants
Mol. Psychiatry
 
2013
18
38
52
5
Owen
M.J.
Craddock
N.
O'Donovan
M.C.
Suggestion of roles for both common and rare risk variants in genome-wide studies of schizophrenia
Arch. Gen. Psychiatry
 
2010
67
667
673
6
Morrow
E.M.
Yoo
S.Y.
Flavell
S.W.
Kim
T.K.
Lin
Y.
Hill
R.S.
Mukaddes
N.M.
Balkhy
S.
Gascon
G.
Hashmi
A.
et al.  
Identifying autism loci and genes by tracing recent shared ancestry
Science
 
2008
321
218
223
7
Lander
E.S.
Botstein
D.
Homozygosity mapping: a way to map human recessive traits with the DNA of inbred children
Science
 
1987
236
1567
1570
8
Gibson
J.
Morton
N.E.
Collins
A.
Extended tracts of homozygosity in outbred human populations
Hum. Mol. Genet.
 
2006
15
789
795
9
Ku
C.S.
Naidoo
N.
Teo
S.M.
Pawitan
Y.
Regions of homozygosity and their impact on complex diseases and traits
Hum. Genet.
 
2011
129
1
15
10
Szpiech
Z.A.
Xu
J.
Pemberton
T.J.
Peng
W.
Zollner
S.
Rosenberg
N.A.
Li
J.Z.
Long runs of homozygosity are enriched for deleterious variation
Am. J. Hum. Genet.
 
2013
93
90
102
11
Lencz
T.
Lambert
C.
DeRosse
P.
Burdick
K.E.
Morgan
T.V.
Kane
J.M.
Kucherlapati
R.
Malhotra
A.K.
Runs of homozygosity reveal highly penetrant recessive loci in schizophrenia
Proc. Natl. Acad. Sci. USA
 
2007
104
19942
19947
12
Keller
M.C.
Simonson
M.A.
Ripke
S.
Neale
B.M.
Gejman
P.V.
Howrigan
D.P.
Lee
S.H.
Lencz
T.
Levinson
D.F.
Sullivan
P.F.
Runs of homozygosity implicate autozygosity as a schizophrenia risk factor
PLoS Genet.
 
2012
8
e1002656
13
Ghani
M.
Sato
C.
Lee
J.H.
Reitz
C.
Moreno
D.
Mayeux
R.
St George-Hyslop
P.
Rogaeva
E.
Evidence of recessive Alzheimer disease loci in a Caribbean Hispanic data set: genome-wide survey of runs of homozygosity
JAMA Neurol.
 
2013
70
1261
1267
14
Yang
H.C.
Chang
L.C.
Liang
Y.J.
Lin
C.H.
Wang
P.L.
A genome-wide homozygosity association study identifies runs of homozygosity associated with rheumatoid arthritis in the human major histocompatibility complex
PloS one
 
2012
7
e34840
15
Casey
J.P.
Magalhaes
T.
Conroy
J.M.
Regan
R.
Shah
N.
Anney
R.
Shields
D.C.
Abrahams
B.S.
Almeida
J.
Bacchelli
E.
et al.  
A novel approach of homozygous haplotype sharing identifies candidate genes in autism spectrum disorder
Hum. Genet.
 
2012
131
565
579
16
Pemberton
T.J.
Absher
D.
Feldman
M.W.
Myers
R.M.
Rosenberg
N.A.
Li
J.Z.
Genomic patterns of homozygosity in worldwide human populations
Am. J. Hum. Genet.
 
2012
91
275
292
17
Kirin
M.
McQuillan
R.
Franklin
C.S.
Campbell
H.
McKeigue
P.M.
Wilson
J.F.
Genomic runs of homozygosity record population history and consanguinity
PloS one
 
2010
5
e13996
18
Gusev
A.
Lowe
J.K.
Stoffel
M.
Daly
M.J.
Altshuler
D.
Breslow
J.L.
Friedman
J.M.
Pe'er
I.
Whole population, genome-wide mapping of hidden relatedness
Genom. Res.
 
2009
19
318
326
19
Browning
B.L.
Browning
S.R.
A fast, powerful method for detecting identity by descent
Am. J. Hum. Genet.
 
2011
88
173
182
20
Gusev
A.
Kenny
E.E.
Lowe
J.K.
Salit
J.
Saxena
R.
Kathiresan
S.
Altshuler
D.M.
Friedman
J.M.
Breslow
J.L.
Pe'er
I.
DASH: a method for identical-by-descent haplotype mapping uncovers association with recent variation
Am. J. Hum. Genet.
 
2011
88
706
717
21
Guha
S.
Rosenfeld
J.A.
Malhotra
A.K.
Lee
A.T.
Gregersen
P.K.
Kane
J.M.
Pe'er
I.
Darvasi
A.
Lencz
T.
Implications for health and disease in the genetic signature of the Ashkenazi Jewish population
Genom. Biol.
 
2012
13
R2
22
Lencz
T.
Guha
S.
Liu
C.
Rosenfeld
J.
Mukherjee
S.
DeRosse
P.
John
M.
Cheng
L.
Zhang
C.
Badner
J.A.
et al.  
Genome-wide association study implicates NDST3 in schizophrenia and bipolar disorder
Nat. Commun.
 
2013
4
2739
23
Ikeda
M.
Aleksic
B.
Kinoshita
Y.
Okochi
T.
Kawashima
K.
Kushima
I.
Ito
Y.
Nakamura
Y.
Kishi
T.
Okumura
T.
et al.  
Genome-wide association study of schizophrenia in a Japanese population
Biol. Psychiat.
 
2011
69
472
478
24
Cazzullo
C.L.
Smeraldi
E.
Penati
G.
The leucocyte antigenic system HL-A as a possible genetic marker of schizophrenia
Br. J. Psychiat.
 
1974
125
25
27
25
Purcell
S.M.
Wray
N.R.
Stone
J.L.
Visscher
P.M.
O'Donovan
M.C.
Sullivan
P.F.
Sklar
P.
Common polygenic variation contributes to risk of schizophrenia and bipolar disorder
Nature
 
2009
460
748
752
26
Shi
J.
Levinson
D.F.
Duan
J.
Sanders
A.R.
Zheng
Y.
Pe'er
I.
Dudbridge
F.
Holmans
P.A.
Whittemore
A.S.
Mowry
B.J.
et al.  
Common variants on chromosome 6p22.1 are associated with schizophrenia
Nature
 
2009
460
753
757
27
Stefansson
H.
Ophoff
R.A.
Steinberg
S.
Andreassen
O.A.
Cichon
S.
Rujescu
D.
Werge
T.
Pietilainen
O.P.
Mors
O.
Mortensen
P.B.
et al.  
Common variants conferring risk of schizophrenia
Nature
 
2009
460
744
747
28
Steinberg
S.
de Jong
S.
Andreassen
O.A.
Werge
T.
Borglum
A.D.
Mors
O.
Mortensen
P.B.
Gustafsson
O.
Costas
J.
Pietilainen
O.P.
et al.  
Common variants at VRK2 and TCF4 conferring risk of schizophrenia
Hum. Mol. Genet.
 
2011
20
4076
4081
29
Tin
A.
Astor
B.C.
Boerwinkle
E.
Hoogeveen
R.C.
Coresh
J.
Kao
W.H.
Genome-wide association study identified the human leukocyte antigen region as a novel locus for plasma beta-2 microglobulin
Hum. Genet.
 
2013
132
619
627
30
Benros
M.E.
Nielsen
P.R.
Nordentoft
M.
Eaton
W.W.
Dalton
S.O.
Mortensen
P.B.
Autoimmune diseases and severe infections as risk factors for schizophrenia: a 30-year population-based register study
Am. J. Psychiat.
 
2011
168
1303
1310
31
Chen
S.J.
Chao
Y.L.
Chen
C.Y.
Chang
C.M.
Wu
E.C.
Wu
C.S.
Yeh
H.H.
Chen
C.H.
Tsai
H.J.
Prevalence of autoimmune diseases in in-patients with schizophrenia: nationwide population-based study
Br. J. Psychiat.
 
2012
200
374
380
32
Eaton
W.W.
Byrne
M.
Ewald
H.
Mors
O.
Chen
C.Y.
Agerbo
E.
Mortensen
P.B.
Association of schizophrenia and autoimmune diseases: linkage of Danish national registers
Am. J. Psychiat.
 
2006
163
521
528
33
Brown
A.S.
Derkits
E.J.
Prenatal infection and schizophrenia: a review of epidemiologic and translational studies
Am. J. Psychiat.
 
2013
167
261
280
34
Khandaker
G.M.
Zimbron
J.
Lewis
G.
Jones
P.B.
Prenatal maternal infection, neurodevelopment and adult schizophrenia: a systematic review of population-based studies
Psychol. Med.
 
2013
43
239
257
35
Brown
A.S.
Patterson
P.H.
Maternal infection and schizophrenia: implications for prevention
Schizop. Bull.
 
2011
37
284
290
36
Sorensen
H.J.
Mortensen
E.L.
Reinisch
J.M.
Mednick
S.A.
Association between prenatal exposure to bacterial infection and risk of schizophrenia
Schizop. Bull.
 
2009
35
631
637
37
Muller
N.
Schwarz
M.J.
Immune system and schizophrenia
Curr. Immun. Rev.
 
2010
6
213
220
38
Saetre
P.
Emilsson
L.
Axelsson
E.
Kreuger
J.
Lindholm
E.
Jazin
E.
Inflammation-related genes up-regulated in schizophrenia brains
BMC Psychiatry
 
2007
7
46
39
Soderlund
J.
Schroder
J.
Nordin
C.
Samuelsson
M.
Walther-Jallow
L.
Karlsson
H.
Erhardt
S.
Engberg
G.
Activation of brain interleukin-1beta in schizophrenia
Mol. Psychiat.
 
2009
14
1069
1071
40
Doherty
P.
Zinkernagel
R.
A biological role for the major histocompatibility antigens
Lancet
 
1975
1
1406
1409
41
Hughes
A.L.
Nei
M.
Pattern of nucleotide substitution at major histocompatibility complex class I loci reveals overdominant selection
Nature
 
1988
335
167
170
42
Spurgin
L.G.
Richardson
D.S.
How pathogens drive genetic diversity: MHC, mechanisms and misunderstandings
Proc. R. Soc. B Biol. Sci
 
2010
277
979
988
43
Palmer
C.G.
Hsieh
H.J.
Reed
E.F.
Lonnqvist
J.
Peltonen
L.
Woodward
J.A.
Sinsheimer
J.S.
HLA-B maternal-fetal genotype matching increases risk of schizophrenia
Am. J. Hum. Genet.
 
2006
79
710
715
44
Kim
Y.
Ripke
S.
Kirov
G.
Sklar
P.
Purcell
S.M.
Owen
M.J.
O'Donovan
M.C.
Sullivan
P.F.
Non-random mating, parent-of-origin, and maternal-fetal incompatibility effects in schizophrenia
Schizop. Res.
 
2013
143
11
17
45
Debnath
M.
Cannon
D.M.
Venkatasubramanian
G.
Variation in the major histocompatibility complex [MHC] gene family in schizophrenia: Associations and functional implications
Prog. Neuro-Psychopharmacol. Biol. Psychiatry
 
2013
42
49
62
46
Cannon
D.M.
Walshe
M.
Dempster
E.
Collier
D.A.
Marshall
N.
Bramon
E.
Murray
R.M.
McDonald
C.
The association of white matter volume in psychotic disorders with genotypic variation in NRG1, MOG and CNP: a voxel-based analysis in affected individuals and their unaffected relatives
Transl. Psychiat.
 
2012
2
e167
47
Liu
X.
Davis
R.L.
The GABAergic anterior paired lateral neuron suppresses and is suppressed by olfactory learning
Nat. Neurosci.
 
2009
12
53
59
48
Sterner
K.N.
Weckle
A.
Chugani
H.T.
Tarca
A.L.
Sherwood
C.C.
Hof
P.R.
Kuzawa
C.W.
Boddy
A.M.
Abbas
A.
Raaum
R.L.
et al.  
Dynamic gene expression in the human cerebral cortex distinguishes children from adults
PloS one
 
2012
7
e37714
49
Agartz
I.
Brown
A.A.
Rimol
L.M.
Hartberg
C.B.
Dale
A.M.
Melle
I.
Djurovic
S.
Andreassen
O.A.
Common sequence variants in the major histocompatibility complex region associate with cerebral ventricular size in schizophrenia
Biol. Psychiat.
 
2011
70
696
698
50
Huh
P.W.
Belayev
L.
Zhao
W.
Koch
S.
Busto
R.
Ginsberg
M.D.
Comparative neuroprotective efficacy of prolonged moderate intraischemic and postischemic hypothermia in focal cerebral ischemia
J. Neurosurg.
 
2000
92
91
99
51
Purcell
S.
Neale
B.
Todd-Brown
K.
Thomas
L.
Ferreira
M.A.
Bender
D.
Maller
J.
Sklar
P.
de Bakker
P.I.
Daly
M.J.
et al.  
PLINK: a tool set for whole-genome association and population-based linkage analyses
Am. J. Hum. Genet.
 
2007
81
559
575
52
Browning
B.L.
Browning
S.R.
A unified approach to genotype imputation and haplotype-phase inference for large data sets of trios and unrelated individuals
Am. J. Hum. Genet.
 
2009
84
210
223
53
Browning
S.R.
Browning
B.L.
Rapid and accurate haplotype phasing and missing-data inference for whole-genome association studies by use of localized haplotype clustering
Am. J. Hum. Genet.
 
2007
81
1084
1097