A simple yet accurate correction for winner's curse can predict signals discovered in much larger genome scans.

MOTIVATION: For genetic studies, statistically significant variants explain far less trait variance than 'sub-threshold' association signals. To dimension follow-up studies, researchers need to accurately estimate 'true' effect sizes at each SNP, e.g. the true mean of odds ratios (ORs)/regression coefficients (RRs) or Z-score noncentralities. Naïve estimates of effect sizes incur winner's curse biases, which are reduced only by laborious winner's curse adjustments (WCAs). Given that Z-scores estimates can be theoretically translated on other scales, we propose a simple method to compute WCA for Z-scores, i.e. their true means/noncentralities. RESULTS: WCA of Z-scores shrinks these towards zero while, on P-value scale, multiple testing adjustment (MTA) shrinks P-values toward one, which corresponds to the zero Z-score value. Thus, WCA on Z-scores scale is a proxy for MTA on P-value scale. Therefore, to estimate Z-score noncentralities for all SNPs in genome scans, we propose F: DR I: nverse Q: uantile T: ransformation (FIQT). It (i) performs the simpler MTA of P-values using FDR and (ii) obtains noncentralities by back-transforming MTA P-values on Z-score scale. When compared to competitors, realistic simulations suggest that FIQT is more (i) accurate and (ii) computationally efficient by orders of magnitude. Practical application of FIQT to Psychiatric Genetic Consortium schizophrenia cohort predicts a non-trivial fraction of sub-threshold signals which become significant in much larger supersamples. CONCLUSIONS: FIQT is a simple, yet accurate, WCA method for Z-scores (and ORs/RRs, via simple transformations). AVAILABILITY AND IMPLEMENTATION: A 10 lines R function implementation is available at https://github.com/bacanusa/FIQT CONTACT: sabacanu@vcu.edu SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

JEPEGMIX: gene-level joint analysis of functional SNPs in cosmopolitan cohorts.

MOTIVATION: To increase detection power, gene level analysis methods are used to aggregate weak signals. To greatly increase computational efficiency, most methods use as input summary statistics from genome-wide association studies (GWAS). Subsequently, gene statistics are constructed using linkage disequilibrium (LD) patterns from a relevant reference panel. However, all methods, including our own Joint Effect on Phenotype of eQTL/functional single nucleotide polymorphisms (SNPs) associated with a Gene (JEPEG), assume homogeneous panels, e.g. European. However, this renders these tools unsuitable for the analysis of large cosmopolitan cohorts. RESULTS: We propose a JEPEG extension, JEPEGMIX, which similar to one of our software tools, Direct Imputation of summary STatistics of unmeasured SNPs from MIXed ethnicity cohorts, is capable of estimating accurate LD patterns for cosmopolitan cohorts. JEPEGMIX uses this accurate LD estimates to (i) impute the summary statistics at unmeasured functional variants and (ii) test for the joint effect of all measured and imputed functional variants which are associated with a gene. We illustrate the performance of our tool by analyzing the GWAS meta-analysis summary statistics from the multi-ethnic Psychiatric Genomics Consortium Schizophrenia stage 2 cohort. This practical application supports the immune system being one of the main drivers of the process leading to schizophrenia. AVAILABILITY AND IMPLEMENTATION: Software, annotation database and examples are available at http://dleelab.github.io/jepegmix/. CONTACT: donghyung.lee@vcuhealth.org SUPPLEMENTARY INFORMATION: Supplementary material is available at Bioinformatics online.

DISTMIX: direct imputation of summary statistics for unmeasured SNPs from mixed ethnicity cohorts.

MOTIVATION: To increase the signal resolution for large-scale meta-analyses of genome-wide association studies, genotypes at unmeasured single nucleotide polymorphisms (SNPs) are commonly imputed using large multi-ethnic reference panels. However, the ever increasing size and ethnic diversity of both reference panels and cohorts makes genotype imputation computationally challenging for moderately sized computer clusters. Moreover, genotype imputation requires subject-level genetic data, which unlike summary statistics provided by virtually all studies, is not publicly available. While there are much less demanding methods which avoid the genotype imputation step by directly imputing SNP statistics, e.g. Directly Imputing summary STatistics (DIST) proposed by our group, their implicit assumptions make them applicable only to ethnically homogeneous cohorts. RESULTS: To decrease computational and access requirements for the analysis of cosmopolitan cohorts, we propose DISTMIX, which extends DIST capabilities to the analysis of mixed ethnicity cohorts. The method uses a relevant reference panel to directly impute unmeasured SNP statistics based only on statistics at measured SNPs and estimated/user-specified ethnic proportions. Simulations show that the proposed method adequately controls the Type I error rates. The 1000 Genomes panel imputation of summary statistics from the ethnically diverse Psychiatric Genetic Consortium Schizophrenia Phase 2 suggests that, when compared to genotype imputation methods, DISTMIX offers comparable imputation accuracy for only a fraction of computational resources. AVAILABILITY AND IMPLEMENTATION: DISTMIX software, its reference population data, and usage examples are publicly available at http://code.google.com/p/distmix. CONTACT: dlee4@vcu.edu SUPPLEMENTARY INFORMATION: Supplementary Data are available at Bioinformatics online.

Genome-wide gene pathway analysis of psychotic illness symptom dimensions based on a new schizophrenia-specific model of the OPCRIT.

Empirically derived phenotypic measurements have the potential to enhance gene-finding efforts in schizophrenia. Previous research based on factor analyses of symptoms has typically included schizoaffective cases. Deriving factor loadings from analysis of only narrowly defined schizophrenia cases could yield more sensitive factor scores for gene pathway and gene ontology analyses. Using an Irish family sample, this study 1) factor analyzed clinician-rated Operational Criteria Checklist items in cases with schizophrenia only, 2) scored the full sample based on these factor loadings, and 3) implemented genome-wide association, gene-based, and gene-pathway analysis of these SCZ-based symptom factors (final N=507). Three factors emerged from the analysis of the schizophrenia cases: a manic, a depressive, and a positive symptom factor. In gene-based analyses of these factors, multiple genes had q<0.01. Of particular interest are findings for PTPRG and WBP1L, both of which were previously implicated by the Psychiatric Genomics Consortium study of SCZ; results from this study suggest that variants in these genes might also act as modifiers of SCZ symptoms. Gene pathway analyses of the first factor indicated over-representation of glutamatergic transmission, GABA-A receptor, and cyclic GMP pathways. Results suggest that these pathways may have differential influence on affective symptom presentation in schizophrenia.

JEPEG: a summary statistics based tool for gene-level joint testing of functional variants.

MOTIVATION: Gene expression is influenced by variants commonly known as expression quantitative trait loci (eQTL). On the basis of this fact, researchers proposed to use eQTL/functional information univariately for prioritizing single nucleotide polymorphisms (SNPs) signals from genome-wide association studies (GWAS). However, most genes are influenced by multiple eQTLs which, thus, jointly affect any downstream phenotype. Therefore, when compared with the univariate prioritization approach, a joint modeling of eQTL action on phenotypes has the potential to substantially increase signal detection power. Nonetheless, a joint eQTL analysis is impeded by (i) not measuring all eQTLs in a gene and/or (ii) lack of access to individual genotypes. RESULTS: We propose joint effect on phenotype of eQTL/functional SNPs associated with a gene (JEPEG), a novel software tool which uses only GWAS summary statistics to (i) impute the summary statistics at unmeasured eQTLs and (ii) test for the joint effect of all measured and imputed eQTLs in a gene. We illustrate the behavior/performance of the developed tool by analysing the GWAS meta-analysis summary statistics from the Psychiatric Genomics Consortium Stage 1 and the Genetic Consortium for Anorexia Nervosa. CONCLUSIONS: Applied analyses results suggest that JEPEG complements commonly used univariate GWAS tools by: (i) increasing signal detection power via uncovering (a) novel genes or (b) known associated genes in smaller cohorts and (ii) assisting in fine-mapping of challenging regions, e.g. major histocompatibility complex for schizophrenia. AVAILABILITY AND IMPLEMENTATION: JEPEG, its associated database of eQTL SNPs and usage examples are publicly available at http://code.google.com/p/jepeg/. CONTACT: dlee4@vcu.edu SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Statistical properties of single-marker tests for rare variants.

With the dramatic technological developments of genome-wide association single-nucleotide polymorphism (SNP) chips and next generation sequencing, human geneticists now have the ability to assay genetic variation at ever-rarer allele frequencies. To fully understand the impact of these rare variants on common, complex diseases, we must be able to accurately assess their statistical significance. However, it is well established that classical association tests are not appropriate for the analysis of low-frequency variation, giving spurious findings when observed counts are too few. To further our understanding of the asymptotic properties of traditional association tests, we conducted a range of simulations of a typical rare variant (~1%) under the null hypothesis and tested the allelic χ2, Cochran-Armitage trend, Wald, and Fisher's exact tests. We demonstrate that rare variation shows marked deviation from the expected distributional behavior for each test, with fewer minor alleles corresponding to a greater degree of test statistics deflation. The effect becomes more pronounced at progressively smaller α levels. We also show that the Wald test is particularly deflated at α levels consistent with genome-wide association significance, much more so than the other association tests considered. In general, these classical association tests are inappropriate for the analysis of variants for which the minor allele is observed fewer than 80 times, largely irrespective of sample size.

Molecular validation of the schizophrenia spectrum.

BACKGROUND: Early descriptive work and controlled family and adoption studies support the hypothesis that a range of personality and nonschizophrenic psychotic disorders aggregate in families of schizophrenic probands. Can we validate, using molecular polygene scores from genome-wide association studies (GWAS), this schizophrenia spectrum? METHODS: The predictive value of polygenic findings reported by the Psychiatric GWAS Consortium (PGC) was applied to 4 groups of relatives from the Irish Study of High-Density Schizophrenia Families (ISHDSF; N = 836) differing on their assignment within the schizophrenia spectrum. Genome-wide single nucleotide polymorphism data for affected and unaffected relatives were used to construct per-individual polygene risk scores based on the PGC stage-I results. We compared mean polygene scores in the ISHDSF with mean scores in ethnically matched population controls (N = 929). RESULTS: The schizophrenia polygene score differed significantly across diagnostic categories and was highest in those with narrow schizophrenia spectrum, lowest in those with no psychiatric illness, and in-between in those classified in the intermediate, broad, and very broad schizophrenia spectrum. Relatives of all of these groups of affected subjects, including those with no diagnosis, had schizophrenia polygene scores significantly higher than the control sample. CONCLUSIONS: In the relatives of high-density families, the observed pattern of enrichment of molecular indices of schizophrenia risk suggests an underlying, continuous liability distribution and validates, using aggregate common risk alleles, a genetic basis for the schizophrenia spectrum disorders. In addition, as predicted by genetic theory, nonpsychotic members of multiply-affected schizophrenia families are significantly enriched for replicated, polygenic risk variants compared with the general population.

Association study of 83 candidate genes for bipolar disorder in chromosome 6q selected using an evidence-based prioritization algorithm.

BACKGROUND: Prior genome-scans of bipolar disorder have revealed chromosome 6q22 as a promising candidate region. However, linkage disequilibrium (LD) mapping studies have yet to identify replicated susceptibility loci. METHODS: We analyzed 1,422 LD-tagging single nucleotide polymorphisms (SNPs) in 83 genes to test single-marker and locus-wide evidence of association with bipolar disorder in the NIMH Genetics Initiative bipolar pedigrees and the Portuguese Island Collection (PIC) (N = 1,093 in 528 informative pairs). Both studies previously demonstrated significant evidence of linkage to 6q. SNPs were genotyped using an Illumina iSelect genotyping array which employs the Infinium assay. Evidence of single-marker association was assessed using the generalized disequilibrium test (GDT). Empirical estimates of gene-wide significance were obtained by permutation (via 100,000 gene-dropping simulations) of Fisher's combined test of P-values for each locus. RESULTS: No single variant yielded significant experiment-wide evidence of association, for either the combined sample or in each subsample. Our gene-dropping simulations identified nominally significant gene-wide associations with multiple loci, of which NT5DC1 in the NIMH subsample and CCNC in the PIC were the strongest candidates. However, no one gene consistently exceeded empirical significance criteria in both independent samples or survived Bonferroni correction for the number of genes tested. CONCLUSIONS: Using a gene-based approach to family-based association, we identified gene-wide associations with several genes, though no single locus was significantly associated with bipolar disorder in both cohorts. This suggests that chromosome 6q may harbor multiple susceptibility loci or that complex patterns of LD in this region may confound approaches based on common SNPs. Published 2013. This article is a U.S. Government work and is in the public domain in the USA.

Association study of 167 candidate genes for schizophrenia selected by a multi-domain evidence-based prioritization algorithm and neurodevelopmental hypothesis.

Integrating evidence from multiple domains is useful in prioritizing disease candidate genes for subsequent testing. We ranked all known human genes (n=3819) under linkage peaks in the Irish Study of High-Density Schizophrenia Families using three different evidence domains: 1) a meta-analysis of microarray gene expression results using the Stanley Brain collection, 2) a schizophrenia protein-protein interaction network, and 3) a systematic literature search. Each gene was assigned a domain-specific p-value and ranked after evaluating the evidence within each domain. For comparison to this ranking process, a large-scale candidate gene hypothesis was also tested by including genes with Gene Ontology terms related to neurodevelopment. Subsequently, genotypes of 3725 SNPs in 167 genes from a custom Illumina iSelect array were used to evaluate the top ranked vs. hypothesis selected genes. Seventy-three genes were both highly ranked and involved in neurodevelopment (category 1) while 42 and 52 genes were exclusive to neurodevelopment (category 2) or highly ranked (category 3), respectively. The most significant associations were observed in genes PRKG1, PRKCE, and CNTN4 but no individual SNPs were significant after correction for multiple testing. Comparison of the approaches showed an excess of significant tests using the hypothesis-driven neurodevelopment category. Random selection of similar sized genes from two independent genome-wide association studies (GWAS) of schizophrenia showed the excess was unlikely by chance. In a further meta-analysis of three GWAS datasets, four candidate SNPs reached nominal significance. Although gene ranking using integrated sources of prior information did not enrich for significant results in the current experiment, gene selection using an a priori hypothesis (neurodevelopment) was superior to random selection. As such, further development of gene ranking strategies using more carefully selected sources of information is warranted.

DIST: direct imputation of summary statistics for unmeasured SNPs.

MOTIVATION: Genotype imputation methods are used to enhance the resolution of genome-wide association studies, and thus increase the detection rate for genetic signals. Although most studies report all univariate summary statistics, many of them limit the access to subject-level genotypes. Because such an access is required by all genotype imputation methods, it is helpful to develop methods that impute summary statistics without going through the interim step of imputing genotypes. Even when subject-level genotypes are available, due to the substantial computational cost of the typical genotype imputation, there is a need for faster imputation methods. RESULTS: Direct Imputation of summary STatistics (DIST) imputes the summary statistics of untyped variants without first imputing their subject-level genotypes. This is achieved by (i) using the conditional expectation formula for multivariate normal variates and (ii) using the correlation structure from a relevant reference population. When compared with genotype imputation methods, DIST (i) requires only a fraction of their computational resources, (ii) has comparable imputation accuracy for independent subjects and (iii) is readily applicable to the imputation of association statistics coming from large pedigree data. Thus, the proposed application is useful for a fast imputation of summary results for (i) studies of unrelated subjects, which (a) do not provide subject-level genotypes or (b) have a large size and (ii) family association studies. AVAILABILITY AND IMPLEMENTATION: Pre-compiled executables built under commonly used operating systems are publicly available at http://code.google.com/p/dist/. CONTACT: dlee4@vcu.edu .

Comprehensive gene-based association study of a chromosome 20 linked region implicates novel risk loci for depressive symptoms in psychotic illness.

BACKGROUND: Prior genomewide scans of schizophrenia support evidence of linkage to regions of chromosome 20. However, association analyses have yet to provide support for any etiologically relevant variants. METHODS: We analyzed 2988 LD-tagging single nucleotide polymorphisms (SNPs) in 327 genes on chromosome 20, to test for association with schizophrenia in 270 Irish high-density families (ISHDSF, N = 270 families, 1408 subjects). These SNPs were genotyped using an Illumina iSelect genotyping array which employs the Infinium assay. Given a previous report of novel linkage with chromosome 20p using latent classes of psychotic illness in this sample, association analysis was also conducted for each of five factor-derived scores based on the Operational Criteria Checklist for Psychotic Illness (delusions, hallucinations, mania, depression, and negative symptoms). Tests of association were conducted using the PDTPHASE and QPDTPHASE packages of UNPHASED. Empirical estimates of gene-wise significance were obtained by adaptive permutation of a) the smallest observed P-value and b) the threshold-truncated product of P-values for each locus. RESULTS: While no single variant was significant after LD-corrected Bonferroni-correction, our gene-dropping analyses identified loci which exceeded empirical significance criteria for both gene-based tests. Namely, R3HDML and C20orf39 are significantly associated with depressive symptoms of schizophrenia (P(emp)<2×10⁻5) based on the minimum P-value and truncated-product methods, respectively. CONCLUSIONS: Using a gene-based approach to family-based association, R3HDML and C20orf39 were found to be significantly associated with clinical dimensions of schizophrenia. These findings demonstrate the efficacy of gene-based analysis and support previous evidence that chromosome 20 may harbor schizophrenia susceptibility or modifier loci.