Genetic variation in 117 myelination-related genes in schizophrenia: Replication of association to lipid biosynthesis genes

Article metrics

Abstract

Schizophrenia is a serious psychotic disorder with high heritability. Several common genetic variants, rare copy number variants and ultra-rare gene-disrupting mutations have been linked to disease susceptibility, but there is still a large gap between the estimated and explained heritability. Since several studies have indicated brain myelination abnormalities in schizophrenia, we aimed to examine whether variants in myelination-related genes could be associated with risk for schizophrenia. We established a set of 117 myelination genes by database searches and manual curation. We used a combination of GWAS (SCZ_N = 35,476; CTRL_N = 46,839), exome chip (SCZ_N = 269; CTRL_N = 336) and exome sequencing data (SCZ_N = 2,527; CTRL_N = 2,536) from schizophrenia cases and healthy controls to examine common and rare variants. We found that a subset of lipid-related genes was nominally associated with schizophrenia (p = 0.037), but this signal did not survive multiple testing correction (FWER = 0.16) and was mainly driven by the SREBF1 and SREBF2 genes that have already been linked to schizophrenia. Further analysis demonstrated that the lowest nominal p-values were p = 0.0018 for a single common variant (rs8539) and p = 0.012 for burden of rare variants (LRP1 gene), but none of them survived multiple testing correction. Our findings suggest that variation in myelination-related genes is not a major risk factor for schizophrenia.

Introduction

Schizophrenia (SCZ) is a serious psychotic disorder with estimated high heritability (0.64–0.81)1,2,3. For a long time, the nature of this genetic susceptibility was not known, but this situation has changed markedly during the last 10 years, especially due to the success of the Psychiatric Genetics Consortium (PGC). So far, 108 gene loci have been significantly associated with SCZ by genome-wide association studies (GWAS)4. The HLA region on chromosome 6 is by far the most significant, including the complement factor 4 locus5. Those common variants account for an estimated 23% of the total heritability, but each variant has only minor influence on the disease risk, with odds ratios typically below 1.103,4. Copy number variants (CNVs) constitute another major type of genetic susceptibility. Such CNVs are rare and may be recurrent (e.g., loci on 1q21.1, 2p16.3 (NRXN1), 3q29, 7q11.2, 15q13.3, 16p11.2 proximal, 16p11.2 distal, and 22q11.2), usually leading to markedly increased risk of schizophrenia in addition to several neuropsychiatric conditions6,7,8,9. Finally, it has been shown that schizophrenia is also associated with inherited, ultra-rare gene-disrupting and putatively protein-damaging variants, mainly in neuron-expressed genes that encode synaptic proteins10.

Despite these important findings, there is still a large gap between the estimated and explained heritability. Alternative approaches are therefore needed to identify more of the genetic factors, and one possibility is to focus on hypotheses driven by the pathophysiology of schizophrenia. Several structural MRI studies including diffusion tensor imaging (DTI) have observed reduced white matter volume and changes in the fibre tracts in patients with schizophrenia11,12,13. Other studies have demonstrated reduced expression of myelination-related genes in post-mortem brain samples14,15,16. These data suggest that myelination abnormalities could be implicated in schizophrenia. Myelin is a multi-lamellar structure that wraps around the axons of the neurons. In the central nervous system (CNS), it is formed by the lipid-rich membrane of oligodendrocytes. The myelination process is complex and temporal, involving specification and proliferation of oligodendrocyte precursor cells (OPCs), differentiation of mature oligodendrocytes, with subsequent synthesis and turnover of structural myelin components including specialized proteins and lipids17,18. Previously, genes implicated in glial functions (astrocytes and oligodendrocytes) have been associated with schizophrenia when tested in gene enrichment analyses19,20,21,22,23,24. The aim of this study was to test more specifically the implication of genes related to myelination as a risk factor for schizophrenia, using a combination of common and rare variants in GWAS, exome chip and exome sequencing data.

Material and Methods

Generation of CNS myelin-related gene set

The CNS myelin-related gene set was generated in a two-step process. First, we queried the Gene Ontology (GO) Consortium database (geneontology.org) by the search terms “myelin” and “myelination”, which gave a large number of GO terms, including “myelination” (436 gene products), “myelin sheath” (214 gene products), “regulation of myelination” (73 gene products), and “central nervous system myelination” (67 gene products). Due to the hierarchical tree structure of the GO database, more specific GO terms with fewer genes were in general contained within the broader terms “myelination” and “myelin sheath”. Since the GO databases may be biased by research focus, and may include both type I and type II errors25,26, the different gene lists were manually curated to select genes with an biologically well-established link to myelin and myelination in the CNS, by careful review of the literature, similar to the approach of Devor et al.27. This enables exclusion of genes with weak documentation of the annotation or functional link to myelin in the peripheral nervous system only. In addition, we included a selection of genes that are involved in lipid biosynthesis and transport (due to the role of lipids in myelination) or in monogenic disorders with structural myelin defects.

The complete CNS myelin-related gene set included a total of 117 genes - 113 protein-coding genes, 2 miRNA, 1 lincRNA and 1 antisense gene (see Supplementary File 1), which was subdivided into genes with biologically known involvement in (1) oligodendrocyte precursor cell (OPC) specification and proliferation (N = 13 genes), (2) oligodendrocyte differentiation and myelination (N = 49 genes; including structural CNS myelin proteins), (3) lipid biosynthesis and transport (N = 29 genes) and (4) various other genes involved in myelination and myelin-related disorders (N = 26 genes).

Clinical samples

The CNS myelin-related gene set was queried in three different samples: (1) a multicenter GWAS sample of 35,476 schizophrenia cases and 46,839 controls from the Psychiatric Genetics Consortium (PGC SCZ2), (2) a Norwegian sample of 269 schizophrenia cases and 336 controls genotyped by the Illumina Exome chip, and (3) a Swedish sample of 2,527 schizophrenia cases and 2,536 controls that were exome sequenced (completely overlapping part of PGC SCZ2 for which we have access to raw data).

PGC schizophrenia GWAS samples

Summary statistics from the PGC-SCZ GWAS were downloaded from https://www.med.unc.edu/pgc/results-and-downloads. The data consist of summary statistics of 9,444,231 imputed variants, on 35,476 cases and 46,839 controls from 46 cohorts of European ancestry (PGC SCZ2 discovery set)4. SNPs with poor imputation (imputation score < 0.9) or minor allele frequencies < 0.1, ambiguous SNPs, and insertions/deletions were filtered out, leaving a total of 3,511,943 markers for analysis.

Norwegian exome chip-genotyped schizophrenia sample

The Norwegian sample consisted of 269 schizophrenia cases and 336 controls (605 unrelated individuals in total) from the TOP study28. All samples had been genotyped with the Illumina HumanExome Beadchip v1.0 array and re-analyzed from raw data using Illumina Genome Studio, according to the best practice from the CHARGE consortium29. Quality of called variants was improved using zCall30. Hardy-Weinberg equilibrium statistics of variants were tested by PLINK31. PLINK/SEQ was applied to transform accepted variants into a vcf file. 1,536 variants (MAF > 0) located within 117 myelination genes were extracted using VCFtools32. In 12 of these variants, the alternative allele was more frequent than the reference allele, so these variants were transposed to match hg19 reference.

Swedish exome sequenced schizophrenia sample

The Swedish sample consisted of 2,527 schizophrenia cases and 2,536 controls (5,063 unrelated individuals in total). All samples were sequenced using either the Agilent SureSelect Human All Exon Kit (29 Mb) or the Agilent SureSelect Human All Exon v.2 Kit (33 Mb) for exome capture33. Sequencing was performed by Illumina GAII or Illumina HiSeq2000. Sequence data were aligned and variants called by the Picard/BWA/GATK pipeline34. On the basis of Exome Chip data and validation of selected variants by Sanger sequencing, a high sensitivity and specificity of singleton calls was estimated as earlier described33. Variant frequencies are available in the dbGaP study phs000473.v1 (http://research.mssm.edu/statgen/sweden). 4,766 variants located within 117 RefSeq defined myelination genes were extracted using VCFtools32. In 46 of these variants, the alternative allele was more frequent than the reference allele, so these variants were transposed to match hg19 reference.

Gene Set Enrichment Analysis (GSEA) in GWAS data

All SNPs in the PGC-SCZ GWAS data were annotated to known NCBI RefSeq genes (date: August 1, 2016)35. The list contained 27,063 genes; in case of several variants, the longest was selected. Since the LD structure of the major histocompatibility complex (MHC) region is challenging, and genes from this region are overrepresented in SCZ GWAS, we excluded genes from hg19 chr6:25–34 Mb, with 26,730 genes remaining. SNPs were assigned to genes if they were a) located within the boundaries of the gene+/−10 kb upstream and downstream or if b) in LD (r2 ≥ 0.20) with another SNP located within the boundaries of the gene (+/−10 kb) using the LDsnpR tool36. For LD estimation, we used the European cohort from the 1000 Genomes Project37.

For each gene, a combined p-value score was calculated according to the Brown method38,39. The technique allows generation of combined p-value scores for dependent tests, where the correction procedure is based on the LD structure within the structural bin. The gene scores were log transformed and ranked in decreasing order.

The Gene Set Enrichment Analysis (GSEA)40 method was developed for global gene expression studies. However, this technique has also been used for GWAS studies41,42. The GSEA performs running sum statistics and calculates an enrichment score for each of the selected myelin-related genes, considering their rank in the pre-ranked gene list from the SCZ GWAS. The significance of the score was calculated using 1,500 permutations of the ranked list.

Analysis of common and rare variants in the exome sequence

Common variants (defined as minor allelic frequency > 0.1) in the 117 myelin-related genes were subject to case control association analysis, using all variants extracted from either the exome chip (Norwegian sample) and exome sequence (Swedish sample) data, as described above. Tests of association were performed using EPACTS single variant Logistic Score Test43, in accordance with the tool documentation.

Rare variants (defined as minor allelic frequency ≤ 0.1) extracted from the Norwegian and Swedish datasets were first analyzed for their potential differential clustering between cases and controls within certain genes (i.e. gene burden analysis), using the SKAT-O test44 according to the EPACTS tool documentation.

We also examined the total number of loci with rare variants per individual (both heterozygous and homozygous, investigated separately) in the total set of 117 genes and the four subsets of genes in the Norwegian and Swedish samples. This analysis method is an approach to extend SKAT-O, which tests variant burden per gene. In contrast to SKAT-O, we test the burden of variants in groups of genes that are related to a certain function. The analysis was performed by reading vcf files into the R/Bioconductor environment, using the VariantAnnotation package45. Comparison of median values of variant counts were used to avoid influence of possible outliers. The difference in variant load between cases and controls was tested by Welch t-test.

Finally, loss of function (LoF) variants in the 117 myelin-related genes in the Swedish sample were annotated and selected using snpEff46 with -lof parameter. Their distribution in cases versus controls was tested with chi2 test in R. We also counted the total number of LoF singleton variants and compared their distribution in cases versus controls.

Results

Lipid biosynthesis genes are associated to schizophrenia in schizophrenia GWAS data

As a first step, we sought to investigate the association of myelin gene groups with schizophrenia in the largest schizophrenia GWAS dataset to date, PGC. The lowest p values for the myelination genes are provided in Supplementary File 1, with the most significant finding being p = 1.806 * 10−13 for rs115370182 (an intron variant in the MOG gene).

By using the publicly available PGC SCZ2 data, we performed a gene set enrichment analysis (GSEA) on the total myelination gene set (117 genes) as well as on the four subsets of genes (see Table 1). There was no significant enrichment for the myelination genes in total. One of the four subsets, the lipid biosynthesis and transport-related genes, demonstrated nominally significant enrichment of association to schizophrenia (p = 0.037), but this signal did not survive correction for multiple testing (FWER = 0.16). Moreover, further analysis of the subset revealed that SREBF1 and SREBF2 were ranked as no. 1 and 2 among the leading edge genes (Supplementary Table 1). Since both genes have previously been associated with schizophrenia47, the subgroup analysis was re-run after exclusion of SREBF1 and SREBF2. The remaining subset of 27 genes failed to display any significant association (p = 0.32). No enrichment of association with schizophrenia was found for the three other gene subsets “OPC specification and proliferation”, “oligodendrocyte differentiation and myelination”, and “other genes involved in myelination and myelin-related disorders” (Table 1).

Table 1 Gene set enrichment analysis (GSEA) for myelin gene sets.

No association with schizophrenia for neither common nor rare variants within the coding sequence of the myelination genes

The next step was to examine association of common variants located in myelination genes to schizophrenia (analysis design provided in Supplementary File 2). We had access to a Norwegian sample of 269 schizophrenia cases and 336 controls (605 unrelated subjects in total) that had been genotyped with the Illumina Human Exome array, and a Swedish sample of 2,527 schizophrenia cases and 2,536 controls (5,063 unrelated individuals in total) that had been exome sequenced. We observed a total of 1,536 and 4,766 different variants within the 113 protein-coding and 4 non-coding myelination-related genes in the Norwegian and Swedish sample, respectively, including 185 and 187 common variants in the two samples (defined as MAF > 0.1). By logistic regression analysis of each common variant, we found no significant differences between cases and controls that survived correction for multiple testing (EPACTS, 5 * 10−8), with the lowest nominal p-value being 1.8 × 10−3 in Swedish data set (Fig. 1 and Table 2, Supplementary Files 3 and 4).

Figure 1
figure1

Manhattan plots for common (MAF > 0.05) variants in Norwegian (A) and Swedish (B) cohorts. Variants located in myelination genes do not reach genome-wide significance.

Table 2 Overview of the distribution of common and rare variants in myelination-related genes in schizophrenia patients and healthy controls.

To test for accumulation of rare variants within a gene, we performed a gene burden analysis with SKAT-O. Differences between cases and controls in the Norwegian (lowest p-value 0.03; EIF4G44) and Swedish samples (lowest p-value 0.012; LPR1) (Table 2) did not survive correction for multiple testing. Detailed information of the SKAT-O gene burden analysis is provided in Supplementary Files 5 and 6.

In theory, a combination of variants in several myelination genes in the same subject could impose increased risk. We therefore examined the load of such variants in cases and controls with respect to the total gene set and each of the four subsets. We measured variant load by calculating cumulative number of heterozygous and homozygous variants. We observed a small difference between cases and controls in the median number of heterozygous variants per subject for all myelination genes in the Swedish sample, but the p-value was not significant after correction for multiple testing (74 in cases versus 75 in controls; t-test p = 0.03; q = 0.30; Fig. 2). For the subset of “other genes involved in myelination and myelin-related disorders” in the Norwegian sample (10 in cases versus 9 in controls; p = 0.007) (Table 2). These differences were minor and none of them were supported by merged data from both samples.

Figure 2
figure2

Distributions of heterozygous and homozygous alternative variant counts in Norwegian and Swedish samples. Boxplots represent first quartile, median and third quartile. Mean values are additionally marked with squares. Each grey dot represents number of myelination gene variants in one subject.

No excess of loss-of-function variants in the myelination genes

In a final step of the analysis, we approached all loss-of-function (LoF) variants, defined as nonsense, frame shift (indels), splice gain and splice loss mutations. Due to a low number of such variants in the Norwegian sample (6 in total), we only used the Swedish dataset that contained 99 different LoF variants across all individuals, including 73 ultra-rare singletons (38 observed in the group of 2,527 schizophrenia patients and 33 among the 2,536 controls; p = 0.55) (Table 2). The remaining 28 non-singleton LoF variants were checked for differences between cases and controls, of which a heterozygous variant in PADI2 (stop gain; allele frequency 0.0015 in gnomAD database) had the lowest p value (chi square test; OR = 0.37, p = 0.03), but the p-value did not survive correction for multiple testing (Table 2).

Discussion

In the current study, we examined the role of common and rare variants in myelination-related genes as risk factors for schizophrenia, using a combination of GWAS, exome chip and exome sequencing data in different case control samples. We observed a nominally significant enrichment of common SNPs in a subset of lipogenesis-related genes in schizophrenia patients (p = 0.037), but conservative correction for multiple testing rejected this finding (FWER = 0.16). Moreover, the trend signal observed in this subset was mainly driven by variants in the well-known schizophrenia-associated genes SREBF1 and SREBF2.

Indeed, SREBF1 and SREBF2, encoding the SREBP1 and SREBP2 transcription factors that control fatty acid and cholesterol biosynthesis, were ranked as the two leading edge genes in the GSEA. Since SREBF1 and SREBF2 on chromosome 17p11.2 and 22q13.2 have both been associated with schizophrenia4,48, the GSEA analysis of the PGC2 schizophrenia GWAS data was re-run after exclusion of these genes. The remaining subset of 27 genes failed to display any significant findings, demonstrating that the signal of association was mainly driven by the SREBF gene loci. It remains to be clarified whether variants in SREBF1 and SREBF2 influence the risk of schizophrenia through effects on lipid biosynthesis and myelination in the brain.

There was no significant enrichment in the PGC2 schizophrenia GWAS data for the total set of myelination genes or for any of the three other subsets of genes, thereby excluding a major role of common SNPs in these genes as disease susceptibility factors. This result is in line with a study by Goudriaan et al.19 who reported an association between schizophrenia and lipid metabolism genes in the PGC1 GWAS data, but no significant findings for a set of 17 myelination genes.

We also examined coding variants in the myelination genes, using exome chip and exome sequencing data. In general, the analyses of single variants, gene burden effects (i.e. number of variants within one gene) and variant load per subject (i.e. accumulation of variants in several genes in one person) did not display any statistically significant associations that survived correction for multiple testing. It is therefore unlikely that coding variants in the myelination-related genes play any major role as risk factors for schizophrenia. This is supported by the recent finding that the increased burden of ultra-rare, gene-disruptive variants in schizophrenia is mostly related to variants in neuronally expressed genes that encode proteins with synaptic localization10. It should be noted that the Swedish exome sequence data used by us overlap with the latter report.

The main strength of our study is the combination of a hypothesis-driven approach, a manually curated set of myelination-related genes and the analysis of both common and rare variants in different samples. The main weakness is the limited statistical power to detect associations between rare variants and schizophrenia, thereby lowering the value of the negative findings.

In conclusion, we have confirmed that non-coding variants linked to the SREBF1 and SREBF2 genes are associated with increased risk of schizophrenia. Our data show that variants located within other myelination related genes are unlikely to be major susceptibility factors in schizophrenia.

References

  1. 1.

    Lichtenstein, P. et al. Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet Lond. Engl. 373, 234–239 (2009).

  2. 2.

    Sniekers, S. et al. Genome-wide association meta-analysis of 78,308 individuals identifies new loci and genes influencing human intelligence. Nat. Genet. 49, 1107–1112 (2017).

  3. 3.

    Ripke, S. et al. Genome-wide association analysis identifies 13 new risk loci for schizophrenia. Nat. Genet. 45, 1150–1159 (2013).

  4. 4.

    Schizophrenia Working Group of the Psychiatric Genomics Consortium. Biological insights from 108 schizophrenia-associated genetic loci. Nature 511 421–427 (2014).

  5. 5.

    Sekar, A. et al. Schizophrenia risk from complex variation of complement component 4. Nature 530, 177–183 (2016).

  6. 6.

    Stefansson, H. et al. Large recurrent microdeletions associated with schizophrenia. Nature 455, 232–236 (2008).

  7. 7.

    Stefansson, H. et al. CNVs conferring risk of autism or schizophrenia affect cognition in controls. Nature 505, 361–366 (2014).

  8. 8.

    CNV and Schizophrenia Working Groups of the Psychiatric Genomics Consortium & Psychosis Endophenotypes International Consortium. Contribution of copy number variants to schizophrenia from a genome-wide study of 41,321 subjects. Nat. Genet. 49, 27–35 (2017).

  9. 9.

    Malhotra, D. & Sebat, J. CNVs: harbingers of a rare variant revolution in psychiatric genetics. Cell 148, 1223–1241 (2012).

  10. 10.

    Genovese, G. et al. Increased burden of ultra-rare protein-altering variants among 4,877 individuals with schizophrenia. Nat. Neurosci. 19, 1433–1441 (2016).

  11. 11.

    Canu, E., Agosta, F. & Filippi, M. A selective review of structural connectivity abnormalities of schizophrenic patients at different stages of the disease. Schizophr. Res. 161, 19–28 (2015).

  12. 12.

    Mighdoll, M. I., Tao, R., Kleinman, J. E. & Hyde, T. M. Myelin, myelin-related disorders, and psychosis. Schizophr. Res. 161, 85–93 (2015).

  13. 13.

    Peters, B. D. & Karlsgodt, K. H. White matter development in the early stages of psychosis. Schizophr. Res. 161, 61–69 (2015).

  14. 14.

    Hakak, Y. et al. Genome-wide expression analysis reveals dysregulation of myelination-related genes in chronic schizophrenia. Proc. Natl. Acad. Sci. USA 98, 4746–4751 (2001).

  15. 15.

    Tkachev, D. et al. Oligodendrocyte dysfunction in schizophrenia and bipolar disorder. Lancet Lond. Engl. 362, 798–805 (2003).

  16. 16.

    Katsel, P., Davis, K. L. & Haroutunian, V. Variations in myelin and oligodendrocyte-related gene expression across multiple brain regions in schizophrenia: a gene ontology study. Schizophr. Res. 79, 157–173 (2005).

  17. 17.

    Yeung, M. S. Y. et al. Dynamics of oligodendrocyte generation and myelination in the human brain. Cell 159, 766–774 (2014).

  18. 18.

    Bercury, K. K. & Macklin, W. B. Dynamics and mechanisms of CNS myelination. Dev. Cell 32, 447–458 (2015).

  19. 19.

    Goudriaan, A. et al. Specific glial functions contribute to schizophrenia susceptibility. Schizophr. Bull. 40, 925–935 (2014).

  20. 20.

    Connor, J. R. et al. Postmortem and imaging based analyses reveal CNS decreased myelination in restless legs syndrome. Sleep Med. 12, 614–619 (2011).

  21. 21.

    Gogtay, N. & Rapoport, J. L. Childhood-onset schizophrenia: insights from neuroimaging studies. J. Am. Acad. Child Adolesc. Psychiatry 47, 1120–1124 (2008).

  22. 22.

    McIntosh, A. M. et al. White matter tractography in bipolar disorder and schizophrenia. Biol. Psychiatry 64, 1088–1092 (2008).

  23. 23.

    Uranova, N. A., Vikhreva, O. V., Rachmanova, V. I. & Orlovskaya, D. D. Ultrastructural alterations of myelinated fibers and oligodendrocytes in the prefrontal cortex in schizophrenia: a postmortem morphometric study. Schizophr. Res. Treat. 2011, 325789 (2011).

  24. 24.

    Uranova, N. A. et al. The role of oligodendrocyte pathology in schizophrenia. Int. J. Neuropsychopharmacol. 10, 537–545 (2007).

  25. 25.

    Schnoes, A. M., Brown, S. D., Dodevski, I. & Babbitt, P. C. Annotation Error in Public Databases: Misannotation of Molecular Function in Enzyme Superfamilies. PLOS Comput. Biol. 5, e1000605 (2009).

  26. 26.

    Huntley, R. P., Sawford, T., Martin, M. J. & O’Donovan, C. Understanding how and why the Gene Ontology and its annotations evolve: the GO within UniProt. GigaScience 3, 4 (2014).

  27. 27.

    Devor, A. et al. Genetic evidence for role of integration of fast and slow neurotransmission in schizophrenia. Mol. Psychiatry 22, 792–801 (2017).

  28. 28.

    Athanasiu, L. et al. Gene variants associated with schizophrenia in a Norwegian genome-wide study are replicated in a large European cohort. J. Psychiatr. Res. 44, 748–753 (2010).

  29. 29.

    Grove, M. L. et al. Best practices and joint calling of the HumanExome BeadChip: the CHARGE Consortium. PloS One 8, e68095 (2013).

  30. 30.

    Goldstein, J. I. et al. zCall: a rare variant caller for array-based genotyping: genetics and population analysis. Bioinforma. Oxf. Engl. 28, 2543–2545 (2012).

  31. 31.

    Purcell, S. et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am. J. Hum. Genet. 81, 559–575 (2007).

  32. 32.

    Danecek, P. et al. The variant call format and VCFtools. Bioinforma. Oxf. Engl. 27, 2156–2158 (2011).

  33. 33.

    Purcell, S. M. et al. A polygenic burden of rare disruptive mutations in schizophrenia. Nature 506, 185–190 (2014).

  34. 34.

    McKenna, A. et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 20, 1297–1303 (2010).

  35. 35.

    O’Leary, N. A. et al. Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Res. 44, D733–745 (2016).

  36. 36.

    Christoforou, A. et al. Linkage-disequilibrium-based binning affects the interpretation of GWASs. Am. J. Hum. Genet. 90, 727–733 (2012).

  37. 37.

    1000 Genomes Project Consortium et al. An integrated map of genetic variation from 1,092 human genomes. Nature 491, 56–65 (2012).

  38. 38.

    Moskvina, V. et al. Evaluation of an approximation method for assessment of overall significance of multiple-dependent tests in a genomewide association study. Genet. Epidemiol. 35, 861–866 (2011).

  39. 39.

    Network and Pathway Analysis Subgroup of Psychiatric Genomics Consortium. Psychiatric genome-wide association study analyses implicate neuronal, immune and histone pathways. Nat. Neurosci. 18 199–209 (2015).

  40. 40.

    Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550 (2005).

  41. 41.

    Zhang, K., Cui, S., Chang, S., Zhang, L. & Wang, J. i-GSEA4GWAS: a web server for identification of pathways/gene sets associated with traits by applying an improved gene set enrichment analysis to genome-wide association study. Nucleic Acids Res. 38, W90–95 (2010).

  42. 42.

    Wang, K., Li, M. & Hakonarson, H. Analysing biological pathways in genome-wide association studies. Nat. Rev. Genet. 11, 843–854 (2010).

  43. 43.

    Lin, D.-Y. & Tang, Z.-Z. A general framework for detecting disease associations with rare variants in sequencing studies. Am. J. Hum. Genet. 89, 354–367 (2011).

  44. 44.

    Lee, S. et al. Optimal unified approach for rare-variant association testing with application to small-sample case-control whole-exome sequencing studies. Am. J. Hum. Genet. 91, 224–237 (2012).

  45. 45.

    Obenchain, V. et al. VariantAnnotation: a Bioconductor package for exploration and annotation of genetic variants. Bioinforma. Oxf. Engl. 30, 2076–2078 (2014).

  46. 46.

    Cingolani, P. et al. A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strainw1118; iso-2; iso-3. Fly (Austin) 6, 80–92 (2012).

  47. 47.

    Le Hellard, S. et al. Association between the insulin-induced gene 2 (INSIG2) and weight gain in a German sample of antipsychotic-treated schizophrenic patients: perturbation of SREBP-controlled lipogenesis in drug-related metabolic adverse effects? Mol. Psychiatry 14, 308–317 (2009).

  48. 48.

    Le Hellard, S. et al. Polymorphisms in SREBF1 and SREBF2, two antipsychotic-activated transcription factors controlling cellular lipogenesis, are associated with schizophrenia in German and Scandinavian samples. Mol. Psychiatry 15, 463–472 (2010).

Download references

Author information

V.S. and T.S. designed and directed the project; S.H., V.S., S.D., C.H., O.A. and P.S. gathered data; T.S., I.S., S.B. and R.K. analysed Norwegian exome chip and Swedish exome sequencing data; T.P. and S.G. analysed large scale GWAS data and performed GSEA calculations; S.H., I.S., R.K., and T.P. provided ideas for statistical computations; T.S. and V.S. wrote the article; S.G. and O.A. helped in results interpretation; P.S., S.B., S.D. and C.H. provided critical revisions of the manuscript, all authors read and accepted final version of the article.

Correspondence to Vidar M. Steen.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.