Advertisement

Novel Functional Genomics Approaches Bridging Neuroscience and Psychiatry

  • Jose M. Restrepo-Lozano
    Affiliations
    Integrated Program in Neuroscience, McGill University, Montreal, Quebec, Canada

    Douglas Mental Health University Institute, Montreal, Quebec, Canada
    Search for articles by this author
  • Cecilia Flores
    Correspondence
    Address correspondence to Cecilia Flores, Ph.D.
    Affiliations
    Department of Neurology and Neurosurgery, Faculty of Medicine, McGill University, Montreal, Quebec, Canada

    Department of Psychiatry, Faculty of Medicine, McGill University, Montreal, Quebec, Canada

    Douglas Mental Health University Institute, Montreal, Quebec, Canada
    Search for articles by this author
  • Patricia P. Silveira
    Correspondence
    Patricia P. Silveira, M.D., Ph.D.
    Affiliations
    Department of Psychiatry, Faculty of Medicine, McGill University, Montreal, Quebec, Canada

    Ludmer Centre for Neuroinformatics and Mental Health, McGill University, Montreal, Quebec, Canada

    Douglas Mental Health University Institute, Montreal, Quebec, Canada
    Search for articles by this author
Open AccessPublished:August 06, 2022DOI:https://doi.org/10.1016/j.bpsgos.2022.07.005

      Abstract

      The possibility of establishing a metric of individual genetic risk for a particular disease or trait has sparked the interest of the clinical and research communities, with many groups developing and validating genomic profiling methodologies for their potential application in clinical care. Current approaches for calculating genetic risk to specific psychiatric conditions consist of aggregating genome-wide association studies–derived estimates into polygenic risk scores, which broadly represent the number of inherited risk alleles for an individual. While the traditional approach for polygenic risk score calculation aggregates estimates of gene-disease associations, novel alternative approaches have started to consider functional molecular phenotypes that are closer to genetic variation and are less penalized by the multiple testing required in genome-wide association studies. Moving the focus from genotype-disease to genotype-gene regulation frameworks, these novel approaches incorporate prior knowledge regarding biological processes involved in disease and aggregate estimates for the association of genotypes and phenotypes using multi-omics data modalities. In this review, we discuss and list different functional genomics tools that can be used and integrated to inform researchers and clinicians for a better understanding and diagnosis of psychopathology. We suggest that these novel approaches can help generate biologically driven hypotheses for polygenic signals that can ultimately serve the clinical community as potential biomarkers of psychiatric disease susceptibility.

      Keywords

      Establishing potential high-risk scenarios prior to the onset of neuropsychiatric conditions could profoundly improve mental health trajectories worldwide by presenting an opportunity for timely interventions, especially during sensitive neurodevelopmental windows. Although the well-established practice of inquiring about an individual’s family history when diagnosing physical and psychiatric conditions is a useful tool to indirectly assess potential heritable risk (
      • Weissman M.M.
      • Wickramaratne P.
      • Adams P.
      • Wolk S.
      • Verdeli H.
      • Olfson M.
      Brief screening for family psychiatric history: The family history screen.
      ,
      • Thompson W.D.
      • Orvaschel H.
      • Prusoff B.A.
      • Kidd K.K.
      An evaluation of the family history method for ascertaining psychiatric disorders.
      ,
      • Baselmans B.M.L.
      • Yengo L.
      • van Rheenen W.
      • Wray N.R.
      Risk in relatives, heritability, SNP-based heritability, and genetic correlations in psychiatric disorders: A review.
      ), an individual’s genomic profile could provide information to guide overall health management. However, the true value of genomic data relies on our understanding of the complex interaction between genes, environments, and lifestyle choices over time (
      • Patel C.J.
      • Ioannidis J.P.
      Studying the elusive environment in large scale.
      ,
      • Marigorta U.M.
      • Gibson G.
      A simulation study of gene-by-environment interactions in GWAS implies ample hidden effects.
      ,
      • Idaghdour Y.
      • Awadalla P.
      Exploiting gene expression variation to capture gene-environment interactions for disease.
      ), and efforts to elucidate this complex interplay have the potential to help develop tools to assess disease susceptibility prior to symptom onset, informing preventive and therapeutic decisions.
      Current genotyping technology allows the identification of inherited DNA differences in the order of millions, mostly in the form of single nucleotide polymorphisms (SNPs), across a given population and in a rapid and affordable manner (
      • Perkel J.
      SNP genotyping: Six technologies that keyed a revolution.
      ). As a result, studying genotype-phenotype associations changed from interrogating a few carefully selected candidate genes at a time to unbiased genome-wide surveys, with constant increases in sample sizes leading to the identification of an increasing number of genetic loci that could modify risk for a given disease (
      • Duncan L.E.
      • Ostacher M.
      • Ballon J.
      How genome-wide association studies (GWAS) made traditional candidate gene studies obsolete.
      ). Although this systematic interrogation of genomes yielded several loci reliably associated with an increased risk for psychiatric phenotypes, linking such loci to specific biological functions remains a challenge, primarily because most identified genome-wide significant associations lie in noncoding portions of the genome and require fine-mapping resolution to determine the real causal variants implicated (
      • Duncan L.E.
      • Ostacher M.
      • Ballon J.
      How genome-wide association studies (GWAS) made traditional candidate gene studies obsolete.
      ,
      • Watanabe K.
      • Stringer S.
      • Frei O.
      • Umićević Mirkov M.
      • de Leeuw C.
      • Polderman T.J.C.
      • et al.
      A global overview of pleiotropy and genetic architecture in complex traits [published correction appears in Nat Genet. 2020 52:353].
      ,
      • Reynolds T.
      • Johnson E.C.
      • Huggett S.B.
      • Bubier J.A.
      • Palmer R.H.C.
      • Agrawal A.
      • et al.
      Interpretation of psychiatric genome-wide association studies with multispecies heterogeneous functional genomic data integration.
      ). Establishing a neurobiological framework underlying psychiatric risk will require a multi-omics data integration approach, with the purpose of mapping the molecular processes linking genomes and disease-relevant phenotypes (
      • Subramanian I.
      • Verma S.
      • Kumar S.
      • Jere A.
      • Anamika K.
      Multi-omics data integration, interpretation, and its application.
      ). Such frameworks may ultimately help improve models of disease risk prediction based on genomic profiles and provide actionable insights for clinical decision making. In this review, we discuss emerging genomic risk assessment approaches in psychiatry, emphasizing methods that explore the neurobiological mechanisms by which gene networks contribute to psychiatric phenotypes.

      Genome-wide Association Studies as the Basis for Mapping Genetic Susceptibility to Psychiatric Phenotypes

      To date, the most common population-based method to find genotype-phenotype associations is the performance of genome-wide association studies (GWASs) [see (
      • Bracken M.B.
      • DeWan A.
      • Hoh J.
      Genome-wide association studies.
      ,
      • Hirschhorn J.N.
      • Daly M.J.
      Genome-wide association studies for common diseases and complex traits.
      )], which has successfully helped identify genomic variants associated with increased risk of developing different psychiatric conditions (
      • Watanabe K.
      • Stringer S.
      • Frei O.
      • Umićević Mirkov M.
      • de Leeuw C.
      • Polderman T.J.C.
      • et al.
      A global overview of pleiotropy and genetic architecture in complex traits [published correction appears in Nat Genet. 2020 52:353].
      ,
      • Buniello A.
      • Macarthur J.A.L.
      • Cerezo M.
      • Harris L.W.
      • Hayhurst J.
      • Malangone C.
      • et al.
      The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics.
      ,
      • Visscher P.M.
      • Wray N.R.
      • Zhang Q.
      • Sklar P.
      • McCarthy M.I.
      • Brown M.A.
      • Yang J.
      10 years of GWAS discovery: Biology, function, and translation.
      ,
      • Wray N.R.
      • Lee S.H.
      • Mehta D.
      • Vinkhuyzen A.A.
      • Dudbridge F.
      • Middeldorp C.M.
      Research Review: Polygenic methods and their application to psychiatric traits.
      ). Essentially, GWASs entail the assessment of millions of variants across many individuals to detect those statistically associated with a specific phenotype. The primary outcome of GWASs typically includes a list of tested variants together with their respective effect sizes. Then, after identifying the relationship between the phenotypic variance and each genotype by means of a linear (for continuous) or logistic (for binary outcomes) regression, significant loci can be functionally annotated for post-GWAS analyses (Figure 1A). Psychiatric genomics studies for conditions such as schizophrenia (
      • Ripke S.
      • Neale B.M.
      • Corvin A.
      • Walters J.T.R.
      • Farh K.-H.
      • Holmans P.A.
      • et al.
      Schizophrenia Working Group of the Psychiatric Genomics Consortium
      iological insights from 108 schizophrenia-associated genetic loci.
      ) and depression (
      • Howard D.M.
      • Adams M.J.
      • Clarke T.K.
      • Hafferty J.D.
      • Gibson J.
      • Shirali M.
      • et al.
      Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions.
      ) have yielded >100 robustly associated risk loci, with ∼43.7% and ∼8.9% of heritability explained by common SNPs, respectively. The remarkable collaborative effort from the Psychiatric Genomics Consortium (PGC) has helped generate important discoveries in the identification of risk-conferring variants as well as in advancing our understanding of the genetic architecture across 11 psychiatric disorders (
      • Ripke S.
      • Neale B.M.
      • Corvin A.
      • Walters J.T.R.
      • Farh K.-H.
      • Holmans P.A.
      • et al.
      Schizophrenia Working Group of the Psychiatric Genomics Consortium
      iological insights from 108 schizophrenia-associated genetic loci.
      ,
      • Howard D.M.
      • Adams M.J.
      • Clarke T.K.
      • Hafferty J.D.
      • Gibson J.
      • Shirali M.
      • et al.
      Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions.
      ,
      • Yu D.
      • Sul J.H.
      • Tsetsos F.
      • Nawaz M.S.
      • Huang A.Y.
      • Zelaya I.
      • et al.
      Interrogating the genetic determinants of Tourette’s syndrome and other Tic disorders through genome-wide association studies.
      ,
      • Grove J.
      • Ripke S.
      • Als T.D.
      • Mattheisen M.
      • Walters R.K.
      • Won H.
      • et al.
      Identification of common genetic risk variants for autism spectrum disorder.
      ,
      • Nievergelt C.M.
      • Maihofer A.X.
      • Klengel T.
      • Atkinson E.G.
      • Chen C.Y.
      • Choi K.W.
      • et al.
      International meta-analysis of PTSD genome-wide association studies identifies sex- and ancestry-specific genetic risk loci.
      ,
      • Watson H.J.
      • Yilmaz Z.
      • Thornton L.M.
      • Hübel C.
      • Coleman J.R.I.
      • Gaspar H.A.
      • et al.
      Genome-wide association study identifies eight risk loci and implicates metabo-psychiatric origins for anorexia nervosa.
      ,
      • Jansen I.E.
      • Savage J.E.
      • Watanabe K.
      • Bryois J.
      • Williams D.M.
      • Steinberg S.
      • et al.
      Genome-wide meta-analysis identifies new loci and functional pathways influencing Alzheimer’s disease risk [published correction appears in Nat Genet. 2020; 52:354.
      ,
      • Pasman J.A.
      • Verweij K.J.H.
      • Gerring Z.
      • Stringer S.
      • Sanchez-Roige S.
      • Treur J.L.
      • et al.
      GWAS of lifetime cannabis use reveals new risk loci, genetic overlap with psychiatric traits, and a causal influence of schizophrenia [published correction appears in Nat Neurosci. 2019; 22:1196].
      ,
      • Walters R.K.
      • Polimanti R.
      • Johnson E.C.
      • McClintick J.N.
      • Adams M.J.
      • Adkins A.E.
      • et al.
      Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
      ,
      • Demontis D.
      • Walters R.K.
      • Martin J.
      • Mattheisen M.
      • Als T.D.
      • Agerbo E.
      • et al.
      Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder.
      ,
      Bipolar Disorder and Schizophrenia Working Group of the Psychiatric Genomics Consortium
      Genomic dissection of bipolar disorder and schizophrenia, including 28 subphenotypes.
      ).
      Figure thumbnail gr1
      Figure 1Overview of the steps involved when conducting a GWAS and creating a PRS. (A) Genotypic data from cases and controls or from a population-based sample are gathered to compare the proportion of specific alleles from each SNP among cases and controls or to determine the linear relationship between genotypes and a continuous trait. After proper quality control of the genotype data and determination of the underlying population structure in the sample, a statistical analysis is conducted to investigate whether the observed allele proportions (for case-control studies) or relationships (for continuous traits) deviate significantly from expected values at each SNP, correcting for the number of tests applied. When an allele is found in the cases more frequently than it would be expected by chance, it is reported as a candidate SNP for the entire haplotype block, together with its estimated effect size which quantifies the increased odds of having the disease per risk allele count. For continuous traits, the regression coefficient will determine the effect size attributed to the “effect” allele. Ideally, the observed GWAS signal should be replicated in an independent cohort to minimize false positives and to calibrate the effect sizes attributed to all SNPs. Genome-wide signals (shown in a Manhattan plot) that have been replicated are typically further investigated during post-GWAS work, which consists of 1) fine-mapping the genomic region to find the true causal variant, 2) investigating the tissues/cell types where the variant is known to be active, 3) determining the genes that are affected by the variant, and 4) identifying the molecular pathways implicated. (B) Using a base and a target dataset, the GWAS-derived estimated effects can be applied to a target sample for which genotype data are available. The calculated PRS is an aggregated score of the individual-level genotype weighted by the SNP effect sizes described in a discovery GWAS, resulting in a normally distributed score in the target sample. The distributions depicted in panel (B) reflect raw standardized values of real PRSs, which could be associated with a particular trait of interest. GWAS, genome-wide association study; OR, odds ratio; PRS, polygenic risk score; QC, quality control; SNP, single nucleotide polymorphism.
      GWAS-derived quantified effects of common human variation have translated into different clinical applications. For example, using data derived from human genetics studies has improved the successful development of novel drugs (
      • Ochoa D.
      • Karim M.
      • Ghoussaini M.
      • Hulcoop D.G.
      • McDonagh E.M.
      • Dunham I.
      Human genetics evidence supports two-thirds of the 2021 FDA-approved drugs.
      ,
      • Nelson M.R.
      • Tipney H.
      • Painter J.L.
      • Shen J.
      • Nicoletti P.
      • Shen Y.
      • et al.
      The support of human genetic evidence for approved drug indications.
      ). Another application central to this review is the calculation of polygenic risk scores (PRSs), which aim to predict the contribution of an individual’s genomic profile to a given trait or disease (
      • Wray N.R.
      • Lee S.H.
      • Mehta D.
      • Vinkhuyzen A.A.
      • Dudbridge F.
      • Middeldorp C.M.
      Research Review: Polygenic methods and their application to psychiatric traits.
      ,
      • Dudbridge F.
      Power and predictive accuracy of polygenic risk scores [published correction appears in PLoS Genet. 2013; 9. doi: 10.1371/annotation/b91ba224-10be-409d-93f4-7423d502cba0].
      ,
      • Wray N.R.
      • Lin T.
      • Austin J.
      • McGrath J.J.
      • Hickie I.B.
      • Murray G.K.
      • et al.
      From basic science to clinical application of polygenic risk scores: A Primer.
      ,
      • Choi S.W.
      • Mak T.S.H.
      • O’Reilly P.F.
      Tutorial: A guide to performing polygenic risk score analyses.
      ,
      • Chatterjee N.
      • Wheeler B.
      • Sampson J.
      • Hartge P.
      • Chanock S.J.
      • Park J.H.
      Projecting the performance of risk prediction based on polygenic analyses of genome-wide association studies.
      ,
      • Torkamani A.
      • Wineinger N.E.
      • Topol E.J.
      The personal and clinical utility of polygenic risk scores.
      ,
      • Lee J.J.
      • Wedow R.
      • Okbay A.
      • Kong E.
      • Maghzian O.
      • Zacher M.
      • et al.
      Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
      ). The possibility of establishing a metric of individual genetic risk for a particular disease or trait has sparked the interest of the clinical community, with many researchers now investigating and exploiting the utility of PRS profiling in clinical care [e.g., (
      • Torkamani A.
      • Wineinger N.E.
      • Topol E.J.
      The personal and clinical utility of polygenic risk scores.
      ) or (
      • Ikeda M.
      • Saito T.
      • Kanazawa T.
      • Iwata N.
      Polygenic risk score as clinical utility in psychiatry: A clinical viewpoint.
      )].

      Aggregating GWAS-Derived Signals Into PRSs: a Proxy for Genetic Liability to Psychiatric Traits

      For many years, studies in psychiatric genetics used a candidate gene approach, investigating the role of SNPs in particular phenotypes [e.g., (
      • Bevilacqua L.
      • Doly S.
      • Kaprio J.
      • Yuan Q.
      • Tikkanen R.
      • Paunio T.
      • et al.
      A population-specific HTR2B stop codon predisposes to severe impulsivity [published correction appears in Nature. 2011; 470:424].
      ), where a specific mutation in the HTR2B gene was associated with increased impulsivity]. However, this approach to study the contribution of common variants to psychiatric phenotypes required a previously defined SNP target that was arbitrarily selected, albeit with very few exceptions. Indeed, conditions such as Huntington’s disease (
      • Caron N.S.
      • Wright G.E.
      • Hayden M.R.
      Huntington Disease.
      ) are caused by large effect variants, and there is a marked increase in risk for Alzheimer’s disease (AD) (although not a determinant of the disease itself) in people with the isoform e4 of the APOE gene (
      • Strittmatter W.J.
      • Saunders A.M.
      • Schmechel D.
      • Pericak-Vance M.
      • Enghild J.
      • Salvesen G.S.
      • et al.
      Apolipoprotein E: High-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease.
      ). However, Huntington’s disease and AD are neurologic conditions with a more defined clinical phenotype compared with psychiatric conditions such as mood disorders, where the degree of polygenicity is even more evident. The candidate gene approach is now considered outdated because it has failed to yield useful insights for psychiatry [see (
      • Duncan L.E.
      • Ostacher M.
      • Ballon J.
      How genome-wide association studies (GWAS) made traditional candidate gene studies obsolete.
      ) for a perspective on how GWASs made candidate gene studies obsolete]. Current psychiatric genetics studies use an unbiased examination of the genome, as a continuously growing body of evidence established the highly polygenic architecture across disorders, with many small-effect risk loci distributed across the entire genome (
      • Gratten J.
      • Wray N.R.
      • Keller M.C.
      • Visscher P.M.
      Large-scale genomics unveils the genetic architecture of psychiatric disorders.
      ,
      • Boyle E.A.
      • Li Y.I.
      • Pritchard J.K.
      An expanded view of complex traits: From polygenic to omnigenic.
      ,
      • Shi H.
      • Kichaev G.
      • Pasaniuc B.
      Contrasting the genetic architecture of 30 complex traits from summary association data.
      ,
      • Zhang Y.
      • Qi G.
      • Park J.H.
      • Chatterjee N.
      Estimation of complex effect-size distributions using summary-level statistics from genome-wide association studies across 32 complex traits.
      ). As psychiatry gradually adopted a more probabilistic and risk-oriented mindset, evidence for a concept that could explain a significant proportion of heritability in independent target samples, based entirely on inherited DNA differences, began to emerge (
      • Purcell S.M.
      • Wray N.R.
      • Stone J.L.
      • Visscher P.M.
      • O’Donovan M.C.
      • et al.
      International Schizophrenia Consortium
      Common polygenic variation contributes to risk of schizophrenia and bipolar disorder.
      ,
      • Wray N.R.
      • Goddard M.E.
      • Visscher P.M.
      Prediction of individual genetic risk of complex disease.
      ).

      Current Methodologies for PRS Calculation in Psychiatry and Important Considerations to Obtain Meaningful Genetic Signals

      In principle, all methods of PRS calculation provide an estimate of an individual’s genetic susceptibility to a trait by aggregating the GWAS-derived effect size estimates into an indexed score, as shown in Figure 1B [for a detailed PRS tutorial, see (
      • Choi S.W.
      • Mak T.S.H.
      • O’Reilly P.F.
      Tutorial: A guide to performing polygenic risk score analyses.
      ); for a detailed PRS review, see (
      • Chatterjee N.
      • Shi J.
      • García-Closas M.
      Developing and evaluating polygenic risk prediction models for stratified disease prevention.
      )]. The classic method of PRS calculation uses clumping or pruning and thresholding (C/P + T method) to prune out SNPs in high linkage disequilibrium and apply varying stringencies to p-value thresholds that can be higher than genome-wide significance to calibrate and maximize predictability (
      • Dudbridge F.
      Power and predictive accuracy of polygenic risk scores [published correction appears in PLoS Genet. 2013; 9. doi: 10.1371/annotation/b91ba224-10be-409d-93f4-7423d502cba0].
      ,
      • Chatterjee N.
      • Shi J.
      • García-Closas M.
      Developing and evaluating polygenic risk prediction models for stratified disease prevention.
      ,
      • Euesden J.
      • Lewis C.M.
      • O’Reilly P.F.
      PRSice: Polygenic Risk Score software.
      ). Essentially, SNPs with p values below an established threshold will keep the original estimate of their effect size, while SNPs with higher p values are excluded from the PRS, shrinking their effect sizes to 0. This process can be carried out iteratively, using a range of p-value thresholds, with the resulting PRSs tested for an association with the target trait in a test sample, determining the optimal p value in a forward selection method (
      • Batra A.
      • Chen L.M.
      • Wang Z.
      • Parent C.
      • Pokhvisneva I.
      • Patel S.
      • et al.
      Early life adversity and polygenic risk for high fasting insulin are associated with childhood impulsivity.
      ,
      • Chen L.M.
      • Tollenaar M.S.
      • Hari Dass S.A.
      • Bouvette-Turcot A.A.
      • Pokhvisneva I.
      • Gaudreau H.
      • et al.
      Maternal antenatal depression and child mental health: Moderation by genomic risk for attention-deficit/hyperactivity disorder.
      ). Other methods for PRS calculation are based on Bayesian frameworks in which the shrinkage of all SNPs is based on a prior distribution specification [for more details, see (
      • Ge T.
      • Chen C.Y.
      • Ni Y.
      • Feng Y.A.
      • Smoller J.W.
      Polygenic prediction via Bayesian regression and continuous shrinkage priors.
      ,
      • Vilhjálmsson B.J.
      • Yang J.
      • Finucane H.K.
      • Gusev A.
      • Lindström S.
      • Ripke S.
      • et al.
      Modeling linkage disequilibrium increases accuracy of polygenic risk scores.
      )]. One example that seems to be particularly suited to calculate PRSs for psychiatric disorders (
      • Ni G.
      • Zeng J.
      • Revez J.A.
      • Wang Y.
      • Zheng Z.
      • Ge T.
      • et al.
      A comparison of ten polygenic score methods for psychiatric disorders applied across multiple cohorts.
      ) is the Bayesian multiple regression summary statistic (SBayesR) (
      • Lloyd-Jones L.R.
      • Zeng J.
      • Sidorenko J.
      • Yengo L.
      • Moser G.
      • Kemper K.E.
      • et al.
      Improved polygenic prediction by Bayesian multiple regression on summary statistics.
      ), which can use publicly available GWAS summary statistics while using prior distributions of alternative genetic effects and analyzing all SNPs together, accounting for their pattern of coinheritance.
      Ideally, a PRS can serve as a tool to stratify the population in terms of disease risk, as this can help decide on potential follow-up actionable measures such as therapeutic interventions, more in-depth screening, or lifestyle modifications. One of the earliest examples of a successful PRS came in 2009 when the International Schizophrenia Consortium (ISC) published an aggregated polygenic signal derived from a GWAS that could predict risk for both schizophrenia and bipolar disorder (
      • Purcell S.M.
      • Wray N.R.
      • Stone J.L.
      • Visscher P.M.
      • O’Donovan M.C.
      • et al.
      International Schizophrenia Consortium
      Common polygenic variation contributes to risk of schizophrenia and bipolar disorder.
      ). As the sample size for the schizophrenia GWASs increased, the phenotypic variance explained by the aggregated polygenic signal also increased. Current estimates indicate that individuals with PRS in the top 10% and top 1% of the population have an approximate 3-fold and 6-fold increase in their risk of developing schizophrenia, respectively, compared with 1% baseline risk when selecting someone randomly from the population (
      • Ripke S.
      • Neale B.M.
      • Corvin A.
      • Walters J.T.R.
      • Farh K.-H.
      • Holmans P.A.
      • et al.
      Schizophrenia Working Group of the Psychiatric Genomics Consortium
      iological insights from 108 schizophrenia-associated genetic loci.
      ,
      • Ripke S.
      PGC SCZ WORKGROUP
      GWAS with over 70.000 cases and 100,000 controls.
      ). Another example comes from the study of Desikan et al. (
      • Desikan R.S.
      • Fan C.C.
      • Wang Y.
      • Schork A.J.
      • Cabral H.J.
      • Cupples L.A.
      • et al.
      Genetic assessment of age-associated Alzheimer disease risk: Development and validation of a polygenic hazard score [published correction appears in PLoS Med. 2017 14:e1002289].
      ), wherein the researchers calculated a PRS based on a large AD GWAS meta-analysis (
      • Lambert J.C.
      • Ibrahim-Verbaas C.A.
      • Harold D.
      • Naj A.C.
      • Sims R.
      • Bellenguez C.
      • et al.
      Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s Disease.
      ) to investigate the PRS predictability of age-specific risk of developing the disease. By combining epidemiological data on population-based incidence rates and PRSs, they found that individuals in the highest PRS quartile developed AD at a lower age and showed the highest yearly AD incidence rate. This finding was then replicated in other independent cohorts, where the PRS was associated with known neurodegenerative markers and with the age of disease onset (
      • Desikan R.S.
      • Fan C.C.
      • Wang Y.
      • Schork A.J.
      • Cabral H.J.
      • Cupples L.A.
      • et al.
      Genetic assessment of age-associated Alzheimer disease risk: Development and validation of a polygenic hazard score [published correction appears in PLoS Med. 2017 14:e1002289].
      ).
      It is important to note that existing GWASs are predominantly performed using individuals of European ancestry. Missing genetic effects present in other populations and genetic variants with very low frequency may dramatically decrease the accuracy of a PRS. This is especially true when the ancestry of the target sample does not match the population of the original GWAS (
      • Peterson R.E.
      • Kuchenbaecker K.
      • Walters R.K.
      • Chen C.Y.
      • Popejoy A.B.
      • Periyasamy S.
      • et al.
      Genome-wide association studies in ancestrally diverse populations: Opportunities, methods, pitfalls, and recommendations.
      ,
      • Martin A.R.
      • Gignoux C.R.
      • Walters R.K.
      • Wojcik G.L.
      • Neale B.M.
      • Gravel S.
      • et al.
      Human demographic history impacts genetic risk prediction across diverse populations [published correction appears in Am J Hum Genet. 2020; 107:788-789].
      ). In addition, it has been shown that PRSs work better when considered in combination with other clinical risk factors, with a joint model improving overall disease risk calculation, the identification of individuals that can benefit from early diagnosis, and predictive accuracy (
      • Desikan R.S.
      • Fan C.C.
      • Wang Y.
      • Schork A.J.
      • Cabral H.J.
      • Cupples L.A.
      • et al.
      Genetic assessment of age-associated Alzheimer disease risk: Development and validation of a polygenic hazard score [published correction appears in PLoS Med. 2017 14:e1002289].
      ,
      • Perkins D.O.
      • Olde Loohuis L.
      • Barbee J.
      • Ford J.
      • Jeffries C.D.
      • Addington J.
      • et al.
      Polygenic risk score contribution to psychosis prediction in a target population of persons at clinical high risk.
      ,
      • Mega J.L.
      • Stitziel N.O.
      • Smith J.G.
      • Chasman D.I.
      • Caulfield M.
      • Devlin J.J.
      • et al.
      Genetic risk, coronary heart disease events, and the clinical benefit of statin therapy: An analysis of primary and secondary prevention trials.
      ,
      • Inouye M.
      • Abraham G.
      • Nelson C.P.
      • Wood A.M.
      • Sweeting M.J.
      • Dudbridge F.
      • et al.
      Genomic risk prediction of coronary artery disease in 480,000 adults: Implications for primary prevention.
      ,
      • Maier R.M.
      • Visscher P.M.
      • Robinson M.R.
      • Wray N.R.
      Embracing polygenicity: A review of methods and tools for psychiatric genetics research.
      ). Prediction is a difficult task, and most GWASs necessitate many millions of individuals to allow PRSs to achieve higher discriminatory power and reach the upper bound of their predictive performance (i.e., heritability estimates) (
      • Chatterjee N.
      • Wheeler B.
      • Sampson J.
      • Hartge P.
      • Chanock S.J.
      • Park J.H.
      Projecting the performance of risk prediction based on polygenic analyses of genome-wide association studies.
      ). Some groups have started to propose alternatives to investigate polygenic signals in psychiatry, considering phenotypes closely linked to genetic variation and therefore more directly affected by it.

      From Genetics to Functional Genomics: PRS Methodologies That Go Beyond the Link Between Genetic Variability and Psychiatric Traits by Addressing Biological Mechanisms/Functions

      The PRS methodologies described so far have been useful tools for clinicians and researchers, but one common characteristic is the agnosticism when it comes to the biological functions implicated in disease risk. In the classic GWAS-PRS methods, the first step consists of identifying statistically significant genetic associations such that afterward, while conducting post-GWAS work, the biological functions implicated in those gene-disease associations can be dissected (Figure 1A) and further explored as potential therapeutic avenues. However, another way to investigate the role that genes play in disease (together with their associated transcripts, proteins, and epigenomes) is to first identify disease-relevant biological processes and functions to create PRSs that somehow capture and quantify those functions and to then test their association with disease (see Table 1). Moving the focus from genotype-disease toward genotype-gene regulation frameworks, below we review these novel methodologies and resources used by some groups to guide the selection of variants and phenotypes, emphasizing those that take into consideration 1) meaningful networks of genes coregulated (or coexpressed) with spatiotemporal specificity and 2) highly quantifiable phenotypes, such as transcriptomic or epigenomic data. We suggest that these approaches can help generate biologically driven hypotheses for polygenic signals that can ultimately serve the clinical community as potential biomarkers for disease susceptibility.
      Table 1Overview of the Different Methodologies for Post-GWAS Analysis
      Method NameDescriptionReference
      MAGMASoftware tool for mapping genome-wide significant variants to genes and gene sets. Novel variations of this method (i.e., H-MAGMA and eMAGMA) are meant to refine the mapping of variants by incorporating long-range and tissue-specific interactions and the enrichment of variants across different gene modules.(
      • de Leeuw C.A.
      • Mooij J.M.
      • Heskes T.
      • Posthuma D.
      MAGMA: Generalized gene-set analysis of GWAS data.
      )
      LDSCMethod that leverages GWAS summary statistics and LD scores from an external panel to distinguish between inflated effect sizes and true polygenic effects. This method is commonly used for determining genetic correlation between complex traits, partitioned heritability, and stratified heritability.(
      • Bulik-Sullivan B.K.
      • Loh P.R.
      • Finucane H.K.
      • Ripke S.
      • Yang J.
      • Schizophrenia Working Group of the Psychiatric Genomics Consortium
      • et al.
      LD Score regression distinguishes confounding from polygenicity in genome-wide association studies.
      )
      SMRMethod that integrates GWAS summary statistics and data from eQTL studies, allowing the user to identify and prioritize genes whose expression levels are associated with specific complex traits.(
      • Zhu Z.
      • Zhang F.
      • Hu H.
      • Bakshi A.
      • Robinson M.R.
      • Powell J.E.
      • et al.
      Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets.
      )
      lassosumMethod to construct a PRS in a penalized regression framework that uses GWAS summary statistics and an LD reference panel.(
      • Mak T.S.H.
      • Porsch R.M.
      • Choi S.W.
      • Zhou X.
      • Sham P.C.
      Polygenic scores via penalized regression on summary statistics.
      )
      LD-HubCentralized database of GWAS summary statistics that automates LDSC analysis pipeline, allowing the user to estimate SNP heritability and genetic correlation across complex traits.(
      • Zheng J.
      • Erzurumluoglu A.M.
      • Elsworth B.L.
      • Kemp J.P.
      • Howe L.
      • Haycock P.C.
      • et al.
      LD Hub: A centralized database and web interface to perform LD score regression that maximizes the potential of summary level GWAS data for SNP heritability and genetic correlation analysis.
      )
      ANNOPREDBayesian framework for disease risk prediction that integrates genomic functional annotations using GWAS summary statistics and estimates LD from reference genotype data.(
      • Hu Y.
      • Lu Q.
      • Powles R.
      • Yao X.
      • Yang C.
      • Fang F.
      • et al.
      Leveraging functional annotations in genetic risk prediction for human complex diseases.
      )
      SBLUPMethod that rescales SNP effect sizes using an external LD reference panel, converting the ordinary least squares SNP estimates into approximate best linear unbiased predictions.(
      • Robinson M.R.
      • Kleinman A.
      • Graff M.
      • Vinkhuyzen A.A.E.
      • Couper D.
      • Miller M.B.
      • et al.
      Genetic evidence of assortative mating in humans.
      )
      PRS-CSPolygenic prediction method that infers posterior effect sizes of SNPs using GWAS summary statistics and an external LD panel. This model places a continuous shrinkage prior on SNP effect sizes.(
      • Ge T.
      • Chen C.Y.
      • Ni Y.
      • Feng Y.A.
      • Smoller J.W.
      Polygenic prediction via Bayesian regression and continuous shrinkage priors.
      )
      JAMPredMethod for modeling polygenic risk using the JAM software, adjusting for local and for long-range LD. The computed polygenic risk predictions are obtained through a Bayesian variable selection framework.(
      • Newcombe P.J.
      • Nelson C.P.
      • Samani N.J.
      • Dudbridge F.
      A flexible and parallelizable approach to genome-wide polygenic risk scores.
      )
      SBayesRPolygenic prediction method that adjusts SNP effect sizes based on Bayesian multiple regression model (BayesR), using GWAS summary statistics data.(
      • Lloyd-Jones L.R.
      • Zeng J.
      • Sidorenko J.
      • Yengo L.
      • Moser G.
      • Kemper K.E.
      • et al.
      Improved polygenic prediction by Bayesian multiple regression on summary statistics.
      )
      LDpred-funcProbabilistic model for deriving PRS that accounts for LD and incorporates trait-specific functional priors to increase prediction accuracy. This model assumes a point-normal distribution as a prior.(
      • Márquez-Luna C.
      • Gazal S.
      • Loh P.R.
      • Kim S.S.
      • Furlotte N.
      • Auton A.
      • et al.
      Incorporating functional priors improves polygenic prediction accuracy in UK Biobank and 23andMe data sets.
      )
      LDpred2Method for deriving polygenic scores using GWAS summary statistics and LD information from an external reference sample to infer posterior mean effect sizes of SNPs. Optimization of LD and p-value thresholds is achieved using a Bayesian framework for shrinkage of SNP effects. This model assumes a point-normal distribution as a prior.(
      • Privé F.
      • Arbel J.
      • Vilhjálmsson B.J.
      LDpred2: Better, faster, stronger [published online ahead of print, 2020].
      )
      PTRSMethod for calculating PTRSs that can be applied as a gene-based complement to other PRS methods, as it does not outperform other current PRS technologies. This method can help improve portability across ancestries and facilitate interpretation of underlying biological effects.(
      • Liang Y.
      • Pividori M.
      • Manichaikul A.
      • Palmer A.A.
      • Cox N.J.
      • Wheeler H.E.
      • et al.
      Polygenic transcriptome risk scores (PTRS) can improve portability of polygenic risk scores across ancestries.
      )
      eQTL, expression quantitative trait locus; GWAS, genome-wide association study; LD, linkage disequilibrium; LDSC, linkage disequilibrium score regression; PRS, polygenic risk score; PTRS, polygenic transcriptome risk score; SNP, single nucleotide polymorphism.
      Genotype-disease effects are small for most common genetic variations, but the fact that a large proportion of disease risk can be explained by variants that modulate gene expression levels (
      • Watanabe K.
      • Stringer S.
      • Frei O.
      • Umićević Mirkov M.
      • de Leeuw C.
      • Polderman T.J.C.
      • et al.
      A global overview of pleiotropy and genetic architecture in complex traits [published correction appears in Nat Genet. 2020 52:353].
      ,
      • Gusev A.
      • Lee S.H.
      • Trynka G.
      • Finucane H.
      • Vilhjálmsson B.J.
      • Xu H.
      • et al.
      Partitioning heritability of regulatory and cell-type-specific variants across 11 common diseases.
      ,
      • Powell S.K.
      • O’Shea C.
      • Brennand K.J.
      • Akbarian S.
      Parsing the functional impact of noncoding genetic variants in the brain epigenome.
      ) is intriguing and may provide clues for the cellular and biological mechanisms underlying disease (
      • Hernandez L.M.
      • Kim M.
      • Hoftman G.D.
      • Haney J.R.
      • de la Torre-Ubieta L.
      • Pasaniuc B.
      • et al.
      Transcriptomic insight into the polygenic mechanisms underlying psychiatric disorders.
      ,
      • Uffelmann E.
      • Posthuma D.
      Emerging methods and resources for biological interrogation of neuropsychiatric polygenic signal.
      ). The transcriptome-wide association study methodology was developed with the goal of detecting associations between measured or predicted levels of gene expression and particular traits (Figure 2A) (
      • Gusev A.
      • Ko A.
      • Shi H.
      • Bhatia G.
      • Chung W.
      • Penninx B.W.
      • et al.
      Integrative approaches for large-scale transcriptome-wide association studies.
      ). For example, in the study of Girgenti et al. (
      • Girgenti M.J.
      • Wang J.
      • Ji D.
      • Cruz D.A.
      • Traumatic Stress Brain Research Group
      • Stein M.B.
      • et al.
      Transcriptomic organization of the human brain in post-traumatic stress disorder.
      ), researchers used the Million Veteran Program posttraumatic stress disorder GWAS dataset to impute gene expression and identify genes significantly associated with posttraumatic stress disorder risk and illness state, uncovering novel functional signals that confer genetic liability for posttraumatic stress disorder. This method provides key advantages with respect to GWAS. First, using a gene-based approach reduces the burden of multiple testing prevalent in other SNP-based approaches. There are approximately 20,000 genes for which one can impute transcript levels. Although large, this number is considerably smaller compared with several million SNPs in a typical GWAS that are individually tested for an association with a given trait. By incorporating functional information about the regulation of gene expression, this method can help uncover the underlying biological mechanisms affecting a trait. Another advantage of this method is that it facilitates the interpretation of the direction of the effect. A gene-based signal that includes the direction of the effect is highly amenable to systems biology approaches because if the increased (or decreased) expression of a gene is associated with a particular trait, the information can be easily incorporated into pathway or network analyses, making the interpretation of results more straightforward, especially when compared with SNP-based signals. This approach is nonetheless limited regarding tissue accessibility in study participants, in particular if the tissue of interest is a specific brain region. To this end, a more novel computational framework such as the probabilistic transcriptome-wide association study can be of help because it can predict gene expression from genotypes and investigate causal relationships between tissue- or cell-type–specific gene expression and complex traits (
      • Zhang Y.
      • Quick C.
      • Yu K.
      • Barbeira A.
      • GTEx Consortium
      • Luca F.
      • et al.
      PTWAS: Investigating tissue-relevant causal molecular mechanisms of complex traits using probabilistic TWAS analysis.
      ).
      Figure thumbnail gr2
      Figure 2Novel approaches to functional genomics. (A) The TWAS consists of associating measured (or predicted) gene expression data with a disease or trait, making this a gene-based rather than a SNP-based association study, considerably reducing the number of multiple comparisons while also providing insight into the potential biological mechanisms driving disease risk. (B) Models that predict gene expression based on genotype data. These models can be tissue- or cell type–specific, allowing testing for the association of the cell- or tissue-specific imputed transcriptome and a given disease or trait. (C) The ePRS model starts with the identification of a gene coexpression network from RNA sequencing data to identify coregulated disease-relevant biological processes in specific tissues. Once a gene network is identified, it maps the genetic variation within the coexpression network in an independent target sample to weigh the SNPs according to a GWAS (typically a GWAS of gene expression, e.g., GTEx). The resulting ePRS, which aims to capture individual variation in the expression of the gene network, can then be associated with a disease or trait, mapping the association between functional biological processes and the target phenotype. (D) A global model that incorporates all the potential intermediate phenotypes that can occur between genetic variation and disease, considering the feedback from the environment and lived experiences across all levels. This model, similar to the ones discussed earlier, generates a unidirectional effect that starts from individual genetic variability and provides alternative approaches to assess the effects of inherited DNA polymorphisms on particular traits. All these approaches can help generate biologically informed predictors of susceptibility to psychiatric-relevant phenotypes. ePRS, expression-based polygenic risk score; GTEx, Genotype-Tissue Expression project; GWAS, genome-wide association study; SNP, single nucleotide polymorphism; TWAS, transcriptome-wide association study.

      Genotype-Based Prediction of Gene Expression in Specific Tissues

      Many research groups are actively developing tools to predict the transcriptional effects of genetic variation [see (
      • Agarwal V.
      • Shendure J.
      Predicting mRNA abundance directly from genomic sequence using deep convolutional neural networks.
      ,
      • Zhou J.
      • Theesfeld C.L.
      • Yao K.
      • Chen K.M.
      • Wong A.K.
      • Troyanskaya O.G.
      Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk.
      ,
      • Alpay B.A.
      • Demetci P.
      • Istrail S.
      • Aguiar D.
      Combinatorial and statistical prediction of gene expression from haplotype sequence.
      ) for some examples], most likely driven by similar motivations: a unidirectional effect (from genes to gene expression) that ultimately narrows the gap between genetic variation and disease. One of the most prominent examples of this type of work is the PrediXcan methodology, which developed a machine learning algorithm to predict tissue-specific gene expression based on genomic profiles (
      • Gamazon E.R.
      • Wheeler H.E.
      • Shah K.P.
      • Mozaffari S.V.
      • Aquino-Michaels K.
      • Carroll R.J.
      • et al.
      A gene-based association method for mapping traits using reference transcriptome data.
      ). Using genotype and gene expression data from the Genotype-Tissue Expression (GTEx) project (
      GTEx Consortium
      The GTEx Consortium atlas of genetic regulatory effects across human tissues.
      ) and other similar datasets, this method generates a database wherein, in a tissue-specific manner, transcript levels can be predicted using as input the genotypic data from any target sample (Figure 2B). PrediXcan serves to calculate an endophenotype (genetically regulated gene expression) that is known to drive biological processes to test for associations with a particular trait [for the entire data repository and the PrediXcan family of methods, see (
      Home, Predict DB. Internet. [cited Dec 27 2021]..
      )]. The more novel version, MultiXcan, can help investigate the mediating role of gene expression on many complex traits, using only summary statistics from publicly available GWASs (
      • Barbeira A.N.
      • Pividori M.
      • Zheng J.
      • Wheeler H.E.
      • Nicolae D.L.
      • Im H.K.
      Integrating predicted transcriptome from multiple tissues improves association detection.
      ).
      The predicted gene expression approach has been applied to existing GWASs for bipolar disorder to identify novel risk-conferring genes PTPRE and BBX, whose predicted transcript levels in whole blood and in the anterior cingulate cortex, respectively, were found to be associated with increased bipolar disorder risk (
      • Shah K.P.
      • Wheeler H.E.
      • Gamazon E.R.
      • Nicolae D.L.
      • Cox N.J.
      • Im H.K.
      Genetic predictors of gene expression associated with risk of bipolar disorder.
      ). This study highlights the importance of gene expression to help understand the potential underlying mechanisms driving disease risk. However, this approach fails to simultaneously consider genes that are coregulated as part of common biological processes, bypassing the established polygenicity of most psychiatric phenotypes.

      Gene Coexpression Networks to Inform Polygenic Metrics

      As discussed previously, most identified genome-wide significant associations are devoid of a clear functional interpretation because they lie in noncoding portions of the genome, requiring fine-mapping resolution to determine the real causal variants implicated (
      • Watanabe K.
      • Stringer S.
      • Frei O.
      • Umićević Mirkov M.
      • de Leeuw C.
      • Polderman T.J.C.
      • et al.
      A global overview of pleiotropy and genetic architecture in complex traits [published correction appears in Nat Genet. 2020 52:353].
      ,
      • Reynolds T.
      • Johnson E.C.
      • Huggett S.B.
      • Bubier J.A.
      • Palmer R.H.C.
      • Agrawal A.
      • et al.
      Interpretation of psychiatric genome-wide association studies with multispecies heterogeneous functional genomic data integration.
      ). Many of these noncoding disease-associated variants are regulatory in nature (a high proportion of these variants have been determined to be cis and/or trans–expression quantitative trait loci [eQTLs]) (
      • Watanabe K.
      • Stringer S.
      • Frei O.
      • Umićević Mirkov M.
      • de Leeuw C.
      • Polderman T.J.C.
      • et al.
      A global overview of pleiotropy and genetic architecture in complex traits [published correction appears in Nat Genet. 2020 52:353].
      ), suggesting that they likely affect the expression of their associated genes, ultimately placing gene expression as an imminent molecular phenotype linking genetics and disease. More crucially, however, disease-associated genes do not operate in isolation, but as part of complex networks that function with an exquisite degree of spatiotemporal specificity for precise biological processes. By operating under the assumption that functional groups of genes are coregulated as part of specific molecular pathways, the identification of disease-relevant and tissue-specific gene networks provides a framework for mapping transcriptionally coregulated processes into a type of polygenic score. This approach can potentially increase the likelihood of discovering psychiatrically relevant markers of disease [see (
      • Hartl C.L.
      • Ramaswami G.
      • Pembroke W.G.
      • Muller S.
      • Pintacuda G.
      • Saha A.
      • et al.
      Coexpression network architecture reveals the brain-wide and multiregional basis of disease susceptibility.
      )].
      A study that aimed to determine genetic susceptibility to cognitive disability used an unsupervised genome-wide coexpression network analysis leveraging measurements of gene expression in human hippocampal tissue, with the goal of capturing modules of covarying genes, which can ultimately provide clues for the molecular mechanisms driving the susceptibility (
      • Johnson M.R.
      • Shkura K.
      • Langley S.R.
      • Delahaye-Duriez A.
      • Srivastava P.
      • Hill W.D.
      • et al.
      Systems genetics identifies a convergent gene network for cognition and neurodevelopmental disease.
      ). The study identified a module of 150 genes with significant enrichment for 1) genes associated with relevant cognitive phenotypes, 2) genes related to neural activity and synaptic processes, and 3) genes intolerant to mutations and that, when mutated, are associated with intellectual disability (
      • Petrovski S.
      • Wang Q.
      • Heinzen E.L.
      • Allen A.S.
      • Goldstein D.B.
      Genic intolerance to functional variation and the interpretation of personal genomes.
      ,
      • Zhu X.
      • Need A.C.
      • Petrovski S.
      • Goldstein D.B.
      One gene, many neuropsychiatric disorders: Lessons from Mendelian diseases.
      ). Another group followed a similar approach, but instead of using an unsupervised analysis, they hypothesized that genes conferring risk to disease must translate into biological risk by acting as part of a coregulated gene network on a measurable molecular phenotype, which could then be associated with the disease (
      • Pergola G.
      • Di Carlo P.
      • D’Ambrosio E.
      • Gelao B.
      • Fazio L.
      • Papalino M.
      • et al.
      DRD2 co-expression network and a related polygenic index predict imaging, behavioral and clinical phenotypes linked to schizophrenia.
      ). They were interested in elucidating the genetic architecture of the D2 receptor molecular pathway because genetic variation within the DRD2 gene has been linked with schizophrenia-related phenotypes, including response to treatment (
      • Ripke S.
      • Neale B.M.
      • Corvin A.
      • Walters J.T.R.
      • Farh K.-H.
      • Holmans P.A.
      • et al.
      Schizophrenia Working Group of the Psychiatric Genomics Consortium
      iological insights from 108 schizophrenia-associated genetic loci.
      ). Starting with human postmortem tissue, the authors identified a prefrontal DRD2 coexpression network using weighted gene coexpression network analysis, and then defined potential SNPs in the form of eQTLs affecting the expression of the genes within the network. Combining these regulatory SNPs into a particular PRS (referred to as polygenic coexpression index), the study captured the genetic component (eQTLs) of the expression of the network and associated the PRS with brain activity measurements during working memory tasks. Finally, they found that individuals with a higher prefrontal cortex DRD2 coexpression PRS are predisposed to a less efficient working memory, which is a known risk-associated phenotype for schizophrenia. This study is an example of how identifying a disease or trait-relevant gene network can help generate hypotheses for novel types of PRSs based on biological frameworks.
      Another innovative way to identify coregulated biological processes underlying the genetic susceptibility to psychiatric conditions leverages data from the GTEx to quantify the genotypic effect linked to gene expression across several tissues. One such example is the method eMAGMA, which integrates both genetic and transcriptomic data to identify disease-specific risk genes and test for their enrichment across different gene modules (

      Gerring ZF, Gamazon ER, Derks EM, Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium (2019): A gene co-expression network-based analysis of multiple brain tissues reveals novel genes and molecular pathways underlying major depression. PLoS Genet 15:e1008245.

      ). This method can exploit a systems biology approach to generate polygenic signals that are essentially based on tissue-specific gene coexpression networks. A similar approach from our group has generated tissue-specific polygenic signals associated with traits or diseases (Figure 2C). We first identify coexpression networks using genome-wide gene expression data from a specific tissue, then map all SNPs within the coexpressed genes and eliminate those in linkage disequilibrium. We then assign to each SNP the weight of the association between alleles and gene expression estimated by GTEx (
      • GTEx Consortium
      • Ardlie K.G.
      • DeLuca D.S.
      • Segrè A.V.
      • Sullivan T.J.
      • Young T.R.
      • et al.
      Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: Multitissue gene regulation in humans.
      ), ultimately obtaining a set of SNPs that lie within a tissue-specific coexpression network, where each SNP is weighted by its estimated influence on gene expression. We can then identify all SNPs from the coexpressed genes in a test sample of subjects with available genotype data and weight the SNPs according to the GTEx. The derived expression-based polygenic signal (or expression-based PRS) reflects variation in the expression of the gene network and can be calculated in target samples with available genotype data (
      • Dalmaz C.
      • Barth B.
      • Pokhvisneva I.
      • Wang Z.
      • Patel S.
      • Quillfeldt J.A.
      • et al.
      Prefrontal cortex VAMP1 gene network moderates the effect of the early environment on cognitive flexibility in children.
      ,
      • de Mendonça Filho E.J.
      • Barth B.
      • Bandeira D.R.
      • de Lima R.M.S.
      • Arcego D.M.
      • Dalmaz C.
      • et al.
      Cognitive development and brain gray matter susceptibility to prenatal adversities: Moderation by the prefrontal cortex brain-derived neurotrophic factor gene co-expression network.
      ,
      • Hari Dass S.A.
      • McCracken K.
      • Pokhvisneva I.
      • Chen L.M.
      • Garg E.
      • Nguyen T.T.T.
      • et al.
      A biologically informed polygenic score identifies endophenotypes and clinical conditions associated with the insulin receptor function on specific brain regions.
      ,
      • Silveira P.P.
      • Pokhvisneva I.
      • Parent C.
      • Cai S.
      • Rema A.S.S.
      • Broekman B.F.P.
      • et al.
      Cumulative prenatal exposure to adversity reveals associations with a broad range of neurodevelopmental outcomes that are moderated by a novel, biologically informed polygenetic score based on the serotonin transporter solute carrier family C6, member 4 (SLC6A4) gene expression.
      ,
      • Miguel P.M.
      • Pereira L.O.
      • Barth B.
      • de Mendonça Filho E.J.
      • Pokhvisneva I.
      • Nguyen T.T.T.
      • et al.
      Prefrontal cortex dopamine transporter gene network moderates the effect of perinatal hypoxic-ischemic conditions on cognitive flexibility and brain gray matter density in children.
      ).
      In a recent study, we investigated whether an expression-based PRS based on corticolimbic-specific gene coexpression networks associates with impulsive phenotypes in children (
      • Restrepo-Lozano J.M.
      • Pokhvisneva I.
      • Wang Z.
      • Patel S.
      • Meaney M.J.
      • Silveira P.P.
      • et al.
      Corticolimbic DCC gene co-expression networks as predictors of impulsivity in children.
      ). We aimed at capturing individual variation in the molecular processes involved in the maturation of corticolimbic substrates, which are known to support inhibitory control. Similar to most studies using functional polygenic signals, we compared the predictive ability of the score against a conventional PRS derived from the latest GWAS for attention-deficit/hyperactivity disorder and found the expression-based PRS to be a better overall predictor of impulsivity. This type of polygenic signal did not have generalizability problems seen in other polygenic score methodologies, as the experiment was conducted in 3 ethnically diverse cohorts, all showing similar effects. This approach exploits the fact that genes engage within complex networks for precise biological functions, and they do so with a remarkable tissue specificity. Based on knowledge of the neurobiological processes of brain development, this score aimed to predict psychiatric-relevant phenotypes.

      Gene-by-Environment Interplay: Quantifying Environmental Influences and Their Interaction With Multi-omics Data

      One of the biggest challenges faced by researchers studying models of disease risk prediction is to develop a methodology to accurately represent an individual’s environment in a quantitative metric. Similar to how a functional PRS can represent a restricted set of phenotype-relevant biological processes, some studies have narrowed down the environment variable to a composite score made up of clearly defined constructs [see (
      • Batra A.
      • Chen L.M.
      • Wang Z.
      • Parent C.
      • Pokhvisneva I.
      • Patel S.
      • et al.
      Early life adversity and polygenic risk for high fasting insulin are associated with childhood impulsivity.
      ,
      • Chen L.M.
      • Tollenaar M.S.
      • Hari Dass S.A.
      • Bouvette-Turcot A.A.
      • Pokhvisneva I.
      • Gaudreau H.
      • et al.
      Maternal antenatal depression and child mental health: Moderation by genomic risk for attention-deficit/hyperactivity disorder.
      ,
      • Dalmaz C.
      • Barth B.
      • Pokhvisneva I.
      • Wang Z.
      • Patel S.
      • Quillfeldt J.A.
      • et al.
      Prefrontal cortex VAMP1 gene network moderates the effect of the early environment on cognitive flexibility in children.
      ,
      • de Mendonça Filho E.J.
      • Barth B.
      • Bandeira D.R.
      • de Lima R.M.S.
      • Arcego D.M.
      • Dalmaz C.
      • et al.
      Cognitive development and brain gray matter susceptibility to prenatal adversities: Moderation by the prefrontal cortex brain-derived neurotrophic factor gene co-expression network.
      ,
      • Silveira P.P.
      • Pokhvisneva I.
      • Parent C.
      • Cai S.
      • Rema A.S.S.
      • Broekman B.F.P.
      • et al.
      Cumulative prenatal exposure to adversity reveals associations with a broad range of neurodevelopmental outcomes that are moderated by a novel, biologically informed polygenetic score based on the serotonin transporter solute carrier family C6, member 4 (SLC6A4) gene expression.
      ,
      • Miguel P.M.
      • Pereira L.O.
      • Barth B.
      • de Mendonça Filho E.J.
      • Pokhvisneva I.
      • Nguyen T.T.T.
      • et al.
      Prefrontal cortex dopamine transporter gene network moderates the effect of perinatal hypoxic-ischemic conditions on cognitive flexibility and brain gray matter density in children.
      ,
      • McGill M.G.
      • Pokhvisneva I.
      • Clappison A.S.
      • McEwen L.M.
      • Beijers R.
      • Tollenaar M.S.
      • et al.
      Maternal prenatal anxiety and the fetal origins of epigenetic aging.
      ) for examples]. By doing so, researchers can start investigating the interplay between genes and environments while also assessing the potential ways in which genetic and environmental effects interact (Figure 2D). Although not in psychiatry, the study by Belsky et al. (
      • Belsky D.W.
      • Domingue B.W.
      • Wedow R.
      • Arseneault L.
      • Boardman J.D.
      • Caspi A.
      • et al.
      Genetic analysis of social-class mobility in five longitudinal studies [published correction appears in Proc Natl Acad Sci U S A. 2018; 115:E10998].
      ) is a good example of how an individual’s environment can exert a powerful influence on his or her socioeconomic attainment. In this study, the authors tested whether a PRS based on a GWAS for educational attainment (which is currently one of the PRSs with highest predictive value) [see (
      • Lee J.J.
      • Wedow R.
      • Okbay A.
      • Kong E.
      • Maghzian O.
      • Zacher M.
      • et al.
      Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
      )] could predict socioeconomic mobility (i.e., any shift in a person’s social class relative to that of their parents). While higher PRSs did predict more socioeconomic success than parents and siblings, additional analyses revealed that the maternal polygenic score is associated with children's educational attainment even when adjusting for the children's own polygenic scores. This suggests an environmentally mediated genetic effect.
      Some studies have started to integrate epigenetic data into genome-wide scores with the goal of identifying individuals who might be at an increased risk of psychiatric phenotypes. For example, given the association between early-life stress and behavioral and psychiatric problems later in life (
      • Lupien S.J.
      • McEwen B.S.
      • Gunnar M.R.
      • Heim C.
      Effects of stress throughout the lifespan on the brain, behaviour and cognition.
      ), the study of Provençal et al. (
      • Provençal N.
      • Arloth J.
      • Cattaneo A.
      • Anacker C.
      • Cattane N.
      • Wiechmann T.
      • et al.
      Glucocorticoid exposure during hippocampal neurogenesis primes future stress response by inducing changes in DNA methylation.
      ) assessed differentially methylated sites following exposure to glucocorticoids in a human hippocampal progenitor cell line and in human blood cells. In addition, a subsequent glucocorticoid exposure induced important transcriptional changes. The overlapping differentially methylated sites were then used to calculate a weighted polyepigenetic score, which was proposed as a potential biomarker for conditions associated with prenatal glucocorticoid exposure in newborns. The calculated score was applied to newborns’ cord blood DNA (n = 817), with the glucocorticoid-responsive score significantly associated with levels of maternal anxiety and depression, suggesting that early-life stress induces lasting epigenetic changes that can ultimately modify the vulnerability to stress exposure in later years.

      Conclusions

      As the field of psychiatric genomics continues to evolve, so will the models of disease risk prediction based on strong biological foundations. Advances in big data availability and complexity (e.g., longitudinal studies such as the ABCD [Adolescent Brain Cognitive Development] cohort, deep phenotyping as in the UK Biobank), mapping the developmental trajectories, and including a wealth of data in large numbers of individuals will benefit the understanding of factors that ultimately play an important role in determining mental health. Polygenic scores and polyepigenetic scores by themselves, like any other marker, have a limited capacity to predict with perfect accuracy the condition for which they were generated. It should be noted, however, that the optimal selection of genetic variants and other genomic markers and the aggregation of their associated weights are active areas of research (
      • Ge T.
      • Chen C.Y.
      • Ni Y.
      • Feng Y.A.
      • Smoller J.W.
      Polygenic prediction via Bayesian regression and continuous shrinkage priors.
      ,
      • Vilhjálmsson B.J.
      • Yang J.
      • Finucane H.K.
      • Gusev A.
      • Lindström S.
      • Ripke S.
      • et al.
      Modeling linkage disequilibrium increases accuracy of polygenic risk scores.
      ,
      • Lloyd-Jones L.R.
      • Zeng J.
      • Sidorenko J.
      • Yengo L.
      • Moser G.
      • Kemper K.E.
      • et al.
      Improved polygenic prediction by Bayesian multiple regression on summary statistics.
      ). The continued improvement of the technology (increases in GWAS sample size and incorporation of different ancestries, higher genotyping resolution, etc.) entails continued revision of the guidelines for their calculation and interpretation. Owing to the recency of the appearance of several methods discussed in this review, evidence of their clinical utility is still lacking, but as the technology driving functional genomics approaches continues to improve, we expect researchers and clinicians to be encouraged to investigate or test their clinical utility in psychiatry. One can assume that some of the methods highlighted here will be replaced by newer approaches. However, incorporating functional aspects rather than being informed exclusively by data-driven approaches is our core message.
      Although it is highlighted that functional PRS can focus on a particular network or system, efforts should be made to maintain a genome-wide platform for unbiased querying of the relevant signals (e.g., genome-wide RNA sequencing). Moreover, one of the advantages of the functional methods is to provide tissue-specific information, but this can be challenging for certain research questions (e.g., epigenetics markers collected from peripheral studies inferring brain mechanisms). Finally, brain gene expression data in humans is postmortem and is limited in numbers, ancestry, and developmental stage representation. These features can influence gene expression and therefore can bias the generation of functional PRS.
      Training of the clinical workforce to handle and communicate genome-wide information is an issue that is becoming more pressing with time. Commercially available and direct-to-consumer genotyping services, which allow users to download their genotype data, are already reporting PRSs for some traits, and users can upload their data into other online PRS calculators (
      • Folkersen L.
      • Pain O.
      • Ingason A.
      • Werge T.
      • Lewis C.M.
      • Austin J.
      Impute.me: An open-source, non-profit tool for using data from direct-to-consumer genetic testing to calculate and interpret polygenic risk scores.
      ). It is important to clearly communicate to the public the utility and, most importantly, the limitations of PRS profiling, in particular driving away the idea that genetic testing can accurately predict every aspect of a person’s health, as it has inherent limitations similar to those of other tests commonly used in clinical settings (
      • Watson J.
      • Jones H.E.
      • Banks J.
      • Whiting P.
      • Salisbury C.
      • Hamilton W.
      Use of multiple inflammatory marker tests in primary care: Using clinical practice research datalink to evaluate accuracy.
      ).
      Future efforts in disease risk prediction should aim at integrating data at multiple levels, aggregating genomics, epigenomics, transcriptomics, proteomics, and metabolomics data into predictive models. Some of the examples presented in this review highlight the significant contribution from each of these data to disease liability. In addition, models should be able to assess the role of the environment at multiple levels (person, family, community) because all biological processes occurring within an individual are physically contained processes functioning together as part of society, including household, neighborhood, school, and work, and so on (
      • Patel C.J.
      • Ioannidis J.P.
      Studying the elusive environment in large scale.
      ). The technological advance should occur in parallel to societal progress in overcoming the menace of racism and structural inequalities, which still are an unfortunate reality that has a major impact on mental and physical health. The ability to combine multilevel biological information with the constant changes in a person’s environment for overall health risk assessment in trusted clinician-patient relationships with joint decision making can revolutionize the diagnosis and early prevention of psychopathology.

      Acknowledgments and Disclosures

      This work was funded by the National Institute on Drug Abuse (Grant No. R01DA037911 [to CF]), the Canadian Institutes of Health Research (Grant No. FRN: 156272 [to CF]; Grant Nos. PJT-166066 and PJT-173237 [to PPS]), the Natural Sciences and Engineering Research Council of Canada (Grant No. RGPIN-2020-04703 [to CF] and Grant No. RGPIN-2018-05063 [to PPS]), a Doctoral Training fellowship from the Fonds de Recherche du Québec en Santé (to JMR-L), and a Healthy Brains for Healthy Lives of the Canada First Research Excellence Fund at McGill University, Canada, Graduate Student Fellowship (to JMR-L).
      This publication is the work of the authors, and Cecilia Flores and Patricia Pelufo Silveira will serve as guarantors for the contents of this article.
      All figures were created using BioRender.com.
      The authors report no biomedical financial interests or potential conflicts of interest.

      References

        • Weissman M.M.
        • Wickramaratne P.
        • Adams P.
        • Wolk S.
        • Verdeli H.
        • Olfson M.
        Brief screening for family psychiatric history: The family history screen.
        Arch Gen Psychiatry. 2000; 57: 675-682
        • Thompson W.D.
        • Orvaschel H.
        • Prusoff B.A.
        • Kidd K.K.
        An evaluation of the family history method for ascertaining psychiatric disorders.
        Arch Gen Psychiatry. 1982; 39: 53-58
        • Baselmans B.M.L.
        • Yengo L.
        • van Rheenen W.
        • Wray N.R.
        Risk in relatives, heritability, SNP-based heritability, and genetic correlations in psychiatric disorders: A review.
        Biol Psychiatry. 2021; 89: 11-19
        • Patel C.J.
        • Ioannidis J.P.
        Studying the elusive environment in large scale.
        JAMA. 2014; 311: 2173-2174
        • Marigorta U.M.
        • Gibson G.
        A simulation study of gene-by-environment interactions in GWAS implies ample hidden effects.
        Front Genet. 2014; 5: 225
        • Idaghdour Y.
        • Awadalla P.
        Exploiting gene expression variation to capture gene-environment interactions for disease.
        Front Genet. 2013; 3: 228
        • Perkel J.
        SNP genotyping: Six technologies that keyed a revolution.
        Nat Methods. 2008; 5: 447-453
        • Duncan L.E.
        • Ostacher M.
        • Ballon J.
        How genome-wide association studies (GWAS) made traditional candidate gene studies obsolete.
        Neuropsychopharmacology. 2019; 44: 1518-1523
        • Watanabe K.
        • Stringer S.
        • Frei O.
        • Umićević Mirkov M.
        • de Leeuw C.
        • Polderman T.J.C.
        • et al.
        A global overview of pleiotropy and genetic architecture in complex traits [published correction appears in Nat Genet. 2020 52:353].
        Nat Genet. 2019; 51: 1339-1348
        • Reynolds T.
        • Johnson E.C.
        • Huggett S.B.
        • Bubier J.A.
        • Palmer R.H.C.
        • Agrawal A.
        • et al.
        Interpretation of psychiatric genome-wide association studies with multispecies heterogeneous functional genomic data integration.
        Neuropsychopharmacology. 2021; 46: 86-97
        • Subramanian I.
        • Verma S.
        • Kumar S.
        • Jere A.
        • Anamika K.
        Multi-omics data integration, interpretation, and its application.
        Bioinform Biol Insights. 2020; 141177932219899051
        • Bracken M.B.
        • DeWan A.
        • Hoh J.
        Genome-wide association studies.
        in: Molecular Epidemiology: Applications in Cancer and Other Human Diseases. CRC Press, Boca Raton2016: 239-252
        • Hirschhorn J.N.
        • Daly M.J.
        Genome-wide association studies for common diseases and complex traits.
        Nat Rev Genet. 2005; 6: 95-108
        • Buniello A.
        • Macarthur J.A.L.
        • Cerezo M.
        • Harris L.W.
        • Hayhurst J.
        • Malangone C.
        • et al.
        The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics.
        Nucleic Acids Res. 2019; 47: D1005-D1012
        • Visscher P.M.
        • Wray N.R.
        • Zhang Q.
        • Sklar P.
        • McCarthy M.I.
        • Brown M.A.
        • Yang J.
        10 years of GWAS discovery: Biology, function, and translation.
        Am J Hum Genet. 2017; 101: 5-22
        • Wray N.R.
        • Lee S.H.
        • Mehta D.
        • Vinkhuyzen A.A.
        • Dudbridge F.
        • Middeldorp C.M.
        Research Review: Polygenic methods and their application to psychiatric traits.
        J Child Psychol Psychiatry. 2014; 55: 1068-1087
        • Ripke S.
        • Neale B.M.
        • Corvin A.
        • Walters J.T.R.
        • Farh K.-H.
        • Holmans P.A.
        • et al.
        • Schizophrenia Working Group of the Psychiatric Genomics Consortium
        iological insights from 108 schizophrenia-associated genetic loci.
        Nat. 2014; 511: 421-427
        • Howard D.M.
        • Adams M.J.
        • Clarke T.K.
        • Hafferty J.D.
        • Gibson J.
        • Shirali M.
        • et al.
        Genome-wide meta-analysis of depression identifies 102 independent variants and highlights the importance of the prefrontal brain regions.
        Nat Neurosci. 2019; 22: 343-352
        • Yu D.
        • Sul J.H.
        • Tsetsos F.
        • Nawaz M.S.
        • Huang A.Y.
        • Zelaya I.
        • et al.
        Interrogating the genetic determinants of Tourette’s syndrome and other Tic disorders through genome-wide association studies.
        Am J Psychiatry. 2019; 176: 217-227
        • Grove J.
        • Ripke S.
        • Als T.D.
        • Mattheisen M.
        • Walters R.K.
        • Won H.
        • et al.
        Identification of common genetic risk variants for autism spectrum disorder.
        Nat Genet. 2019; 51: 431-444
        • Nievergelt C.M.
        • Maihofer A.X.
        • Klengel T.
        • Atkinson E.G.
        • Chen C.Y.
        • Choi K.W.
        • et al.
        International meta-analysis of PTSD genome-wide association studies identifies sex- and ancestry-specific genetic risk loci.
        Nat Commun. 2019; 10: 4558
        • Watson H.J.
        • Yilmaz Z.
        • Thornton L.M.
        • Hübel C.
        • Coleman J.R.I.
        • Gaspar H.A.
        • et al.
        Genome-wide association study identifies eight risk loci and implicates metabo-psychiatric origins for anorexia nervosa.
        Nat Genet. 2019; 51: 1207-1214
        • Jansen I.E.
        • Savage J.E.
        • Watanabe K.
        • Bryois J.
        • Williams D.M.
        • Steinberg S.
        • et al.
        Genome-wide meta-analysis identifies new loci and functional pathways influencing Alzheimer’s disease risk [published correction appears in Nat Genet. 2020; 52:354.
        Nat Genet. 2019; 51: 404-413
        • Pasman J.A.
        • Verweij K.J.H.
        • Gerring Z.
        • Stringer S.
        • Sanchez-Roige S.
        • Treur J.L.
        • et al.
        GWAS of lifetime cannabis use reveals new risk loci, genetic overlap with psychiatric traits, and a causal influence of schizophrenia [published correction appears in Nat Neurosci. 2019; 22:1196].
        Nat Neurosci. 2018; 21: 1161-1170
        • Walters R.K.
        • Polimanti R.
        • Johnson E.C.
        • McClintick J.N.
        • Adams M.J.
        • Adkins A.E.
        • et al.
        Transancestral GWAS of alcohol dependence reveals common genetic underpinnings with psychiatric disorders.
        Nat Neurosci. 2018; 21: 1656-1669
        • Demontis D.
        • Walters R.K.
        • Martin J.
        • Mattheisen M.
        • Als T.D.
        • Agerbo E.
        • et al.
        Discovery of the first genome-wide significant risk loci for attention deficit/hyperactivity disorder.
        Nat Genet. 2019; 51: 63-75
        • Bipolar Disorder and Schizophrenia Working Group of the Psychiatric Genomics Consortium
        Genomic dissection of bipolar disorder and schizophrenia, including 28 subphenotypes.
        Cell. 2018; 173: 1705-1715.e16
        • Ochoa D.
        • Karim M.
        • Ghoussaini M.
        • Hulcoop D.G.
        • McDonagh E.M.
        • Dunham I.
        Human genetics evidence supports two-thirds of the 2021 FDA-approved drugs.
        Nat Rev Drug Discov. 2022; 21: 551
        • Nelson M.R.
        • Tipney H.
        • Painter J.L.
        • Shen J.
        • Nicoletti P.
        • Shen Y.
        • et al.
        The support of human genetic evidence for approved drug indications.
        Nat Genet. 2015; 47: 856-860
        • Dudbridge F.
        Power and predictive accuracy of polygenic risk scores [published correction appears in PLoS Genet. 2013; 9. doi: 10.1371/annotation/b91ba224-10be-409d-93f4-7423d502cba0].
        PLoS Genet. 2013; 9e1003348
        • Wray N.R.
        • Lin T.
        • Austin J.
        • McGrath J.J.
        • Hickie I.B.
        • Murray G.K.
        • et al.
        From basic science to clinical application of polygenic risk scores: A Primer.
        JAMA Psychiatry. 2021; 78: 101-109
        • Choi S.W.
        • Mak T.S.H.
        • O’Reilly P.F.
        Tutorial: A guide to performing polygenic risk score analyses.
        Nat Protoc. 2020; 15: 2759-2772
        • Chatterjee N.
        • Wheeler B.
        • Sampson J.
        • Hartge P.
        • Chanock S.J.
        • Park J.H.
        Projecting the performance of risk prediction based on polygenic analyses of genome-wide association studies.
        Nat Genet. 2013; 45: 400-405e1--3
        • Torkamani A.
        • Wineinger N.E.
        • Topol E.J.
        The personal and clinical utility of polygenic risk scores.
        Nat Rev Genet. 2018; 19: 581-590
        • Lee J.J.
        • Wedow R.
        • Okbay A.
        • Kong E.
        • Maghzian O.
        • Zacher M.
        • et al.
        Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals.
        Nat Genet. 2018; 50: 1112-1121
        • Ikeda M.
        • Saito T.
        • Kanazawa T.
        • Iwata N.
        Polygenic risk score as clinical utility in psychiatry: A clinical viewpoint.
        J Hum Genet. 2021; 66: 53-60
        • Bevilacqua L.
        • Doly S.
        • Kaprio J.
        • Yuan Q.
        • Tikkanen R.
        • Paunio T.
        • et al.
        A population-specific HTR2B stop codon predisposes to severe impulsivity [published correction appears in Nature. 2011; 470:424].
        Nature. 2010; 468: 1061-1066
        • Caron N.S.
        • Wright G.E.
        • Hayden M.R.
        Huntington Disease.
        in: Adam M.P. Mirzaa G.M. Pagon R.A. University of Washington, Seattle, GeneReviews. Seattle (WA)October 23, 1998 (updated 2020 Jun 11)
        • Strittmatter W.J.
        • Saunders A.M.
        • Schmechel D.
        • Pericak-Vance M.
        • Enghild J.
        • Salvesen G.S.
        • et al.
        Apolipoprotein E: High-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease.
        Proc Natl Acad Sci U S A. 1993; 90: 1977-1981
        • Gratten J.
        • Wray N.R.
        • Keller M.C.
        • Visscher P.M.
        Large-scale genomics unveils the genetic architecture of psychiatric disorders.
        Nat Neurosci. 2014; 17: 782-790
        • Boyle E.A.
        • Li Y.I.
        • Pritchard J.K.
        An expanded view of complex traits: From polygenic to omnigenic.
        Cell. 2017; 169: 1177-1186
        • Shi H.
        • Kichaev G.
        • Pasaniuc B.
        Contrasting the genetic architecture of 30 complex traits from summary association data.
        Am J Hum Genet. 2016; 99: 139-153
        • Zhang Y.
        • Qi G.
        • Park J.H.
        • Chatterjee N.
        Estimation of complex effect-size distributions using summary-level statistics from genome-wide association studies across 32 complex traits.
        Nat Genet. 2018; 50: 1318-1326
        • Purcell S.M.
        • Wray N.R.
        • Stone J.L.
        • Visscher P.M.
        • O’Donovan M.C.
        • et al.
        • International Schizophrenia Consortium
        Common polygenic variation contributes to risk of schizophrenia and bipolar disorder.
        Nature. 2009; 460: 748-752
        • Wray N.R.
        • Goddard M.E.
        • Visscher P.M.
        Prediction of individual genetic risk of complex disease.
        Curr Opin Genet Dev. 2008; 18: 257-263
        • Chatterjee N.
        • Shi J.
        • García-Closas M.
        Developing and evaluating polygenic risk prediction models for stratified disease prevention.
        Nat Rev Genet. 2016; 17: 392-406
        • Euesden J.
        • Lewis C.M.
        • O’Reilly P.F.
        PRSice: Polygenic Risk Score software.
        Bioinformatics. 2015; 31: 1466-1468
        • Batra A.
        • Chen L.M.
        • Wang Z.
        • Parent C.
        • Pokhvisneva I.
        • Patel S.
        • et al.
        Early life adversity and polygenic risk for high fasting insulin are associated with childhood impulsivity.
        Front Neurosci. 2021; 15704785
        • Chen L.M.
        • Tollenaar M.S.
        • Hari Dass S.A.
        • Bouvette-Turcot A.A.
        • Pokhvisneva I.
        • Gaudreau H.
        • et al.
        Maternal antenatal depression and child mental health: Moderation by genomic risk for attention-deficit/hyperactivity disorder.
        Dev Psychopathol. 2020; 32: 1810-1821
        • Ge T.
        • Chen C.Y.
        • Ni Y.
        • Feng Y.A.
        • Smoller J.W.
        Polygenic prediction via Bayesian regression and continuous shrinkage priors.
        Nat Commun. 2019; 10: 1776
        • Vilhjálmsson B.J.
        • Yang J.
        • Finucane H.K.
        • Gusev A.
        • Lindström S.
        • Ripke S.
        • et al.
        Modeling linkage disequilibrium increases accuracy of polygenic risk scores.
        Am J Hum Genet. 2015; 97: 576-592
        • Ni G.
        • Zeng J.
        • Revez J.A.
        • Wang Y.
        • Zheng Z.
        • Ge T.
        • et al.
        A comparison of ten polygenic score methods for psychiatric disorders applied across multiple cohorts.
        Biol Psychiatry. 2021; 90: 611-620
        • Lloyd-Jones L.R.
        • Zeng J.
        • Sidorenko J.
        • Yengo L.
        • Moser G.
        • Kemper K.E.
        • et al.
        Improved polygenic prediction by Bayesian multiple regression on summary statistics.
        Nat Commun. 2019; 10: 5086
        • Ripke S.
        • PGC SCZ WORKGROUP
        GWAS with over 70.000 cases and 100,000 controls.
        Eur Neuropsychopharmacol. 2019; 29: S814
        • Desikan R.S.
        • Fan C.C.
        • Wang Y.
        • Schork A.J.
        • Cabral H.J.
        • Cupples L.A.
        • et al.
        Genetic assessment of age-associated Alzheimer disease risk: Development and validation of a polygenic hazard score [published correction appears in PLoS Med. 2017 14:e1002289].
        PLoS Med. 2017; 14e1002258
        • Lambert J.C.
        • Ibrahim-Verbaas C.A.
        • Harold D.
        • Naj A.C.
        • Sims R.
        • Bellenguez C.
        • et al.
        Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s Disease.
        Nat Genet. 2013; 45: 1452-1458
        • Peterson R.E.
        • Kuchenbaecker K.
        • Walters R.K.
        • Chen C.Y.
        • Popejoy A.B.
        • Periyasamy S.
        • et al.
        Genome-wide association studies in ancestrally diverse populations: Opportunities, methods, pitfalls, and recommendations.
        Cell. 2019; 179: 589-603
        • Martin A.R.
        • Gignoux C.R.
        • Walters R.K.
        • Wojcik G.L.
        • Neale B.M.
        • Gravel S.
        • et al.
        Human demographic history impacts genetic risk prediction across diverse populations [published correction appears in Am J Hum Genet. 2020; 107:788-789].
        Am J Hum Genet. 2017; 100: 635-649
        • Perkins D.O.
        • Olde Loohuis L.
        • Barbee J.
        • Ford J.
        • Jeffries C.D.
        • Addington J.
        • et al.
        Polygenic risk score contribution to psychosis prediction in a target population of persons at clinical high risk.
        Am J Psychiatry. 2020; 177: 155-163
        • Mega J.L.
        • Stitziel N.O.
        • Smith J.G.
        • Chasman D.I.
        • Caulfield M.
        • Devlin J.J.
        • et al.
        Genetic risk, coronary heart disease events, and the clinical benefit of statin therapy: An analysis of primary and secondary prevention trials.
        Lancet. 2015; 385: 2264-2271
        • Inouye M.
        • Abraham G.
        • Nelson C.P.
        • Wood A.M.
        • Sweeting M.J.
        • Dudbridge F.
        • et al.
        Genomic risk prediction of coronary artery disease in 480,000 adults: Implications for primary prevention.
        J Am Coll Cardiol. 2018; 72: 1883-1893
        • Maier R.M.
        • Visscher P.M.
        • Robinson M.R.
        • Wray N.R.
        Embracing polygenicity: A review of methods and tools for psychiatric genetics research.
        Psychol Med. 2018; 48: 1055-1067
        • Gusev A.
        • Lee S.H.
        • Trynka G.
        • Finucane H.
        • Vilhjálmsson B.J.
        • Xu H.
        • et al.
        Partitioning heritability of regulatory and cell-type-specific variants across 11 common diseases.
        Am J Hum Genet. 2014; 95: 535-552
        • Powell S.K.
        • O’Shea C.
        • Brennand K.J.
        • Akbarian S.
        Parsing the functional impact of noncoding genetic variants in the brain epigenome.
        Biol Psychiatry. 2021; 89: 65-75
        • Hernandez L.M.
        • Kim M.
        • Hoftman G.D.
        • Haney J.R.
        • de la Torre-Ubieta L.
        • Pasaniuc B.
        • et al.
        Transcriptomic insight into the polygenic mechanisms underlying psychiatric disorders.
        Biol Psychiatry. 2021; 89: 54-64
        • Uffelmann E.
        • Posthuma D.
        Emerging methods and resources for biological interrogation of neuropsychiatric polygenic signal.
        Biol Psychiatry. 2021; 89: 41-53
        • Gusev A.
        • Ko A.
        • Shi H.
        • Bhatia G.
        • Chung W.
        • Penninx B.W.
        • et al.
        Integrative approaches for large-scale transcriptome-wide association studies.
        Nat Genet. 2016; 48: 245-252
        • Girgenti M.J.
        • Wang J.
        • Ji D.
        • Cruz D.A.
        • Traumatic Stress Brain Research Group
        • Stein M.B.
        • et al.
        Transcriptomic organization of the human brain in post-traumatic stress disorder.
        Nat Neurosci. 2021; 24: 24-33
        • Zhang Y.
        • Quick C.
        • Yu K.
        • Barbeira A.
        • GTEx Consortium
        • Luca F.
        • et al.
        PTWAS: Investigating tissue-relevant causal molecular mechanisms of complex traits using probabilistic TWAS analysis.
        Genome Biol. 2020; 21: 232
        • Agarwal V.
        • Shendure J.
        Predicting mRNA abundance directly from genomic sequence using deep convolutional neural networks.
        Cell Rep. 2020; 31107663
        • Zhou J.
        • Theesfeld C.L.
        • Yao K.
        • Chen K.M.
        • Wong A.K.
        • Troyanskaya O.G.
        Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk.
        Nat Genet. 2018; 50: 1171-1179
        • Alpay B.A.
        • Demetci P.
        • Istrail S.
        • Aguiar D.
        Combinatorial and statistical prediction of gene expression from haplotype sequence.
        Bioinformatics. 2020; 36: i194-202
        • Gamazon E.R.
        • Wheeler H.E.
        • Shah K.P.
        • Mozaffari S.V.
        • Aquino-Michaels K.
        • Carroll R.J.
        • et al.
        A gene-based association method for mapping traits using reference transcriptome data.
        Nat Genet. 2015; 47: 1091-1098
        • GTEx Consortium
        The GTEx Consortium atlas of genetic regulatory effects across human tissues.
        Science. 2020; 369: 1318-1330
      1. Home, Predict DB. Internet. [cited Dec 27 2021]..
        (Available at:)
        https://predictdb.org/
        Date accessed: December 27, 2021
        • Barbeira A.N.
        • Pividori M.
        • Zheng J.
        • Wheeler H.E.
        • Nicolae D.L.
        • Im H.K.
        Integrating predicted transcriptome from multiple tissues improves association detection.
        PLoS Genet. 2019; 15e1007889
        • Shah K.P.
        • Wheeler H.E.
        • Gamazon E.R.
        • Nicolae D.L.
        • Cox N.J.
        • Im H.K.
        Genetic predictors of gene expression associated with risk of bipolar disorder.
        bioRxiv. 2016; 043752
        • Hartl C.L.
        • Ramaswami G.
        • Pembroke W.G.
        • Muller S.
        • Pintacuda G.
        • Saha A.
        • et al.
        Coexpression network architecture reveals the brain-wide and multiregional basis of disease susceptibility.
        Nat Neurosci. 2021; 24: 1313-1323
        • Johnson M.R.
        • Shkura K.
        • Langley S.R.
        • Delahaye-Duriez A.
        • Srivastava P.
        • Hill W.D.
        • et al.
        Systems genetics identifies a convergent gene network for cognition and neurodevelopmental disease.
        Nat Neurosci. 2016; 19: 223-232
        • Petrovski S.
        • Wang Q.
        • Heinzen E.L.
        • Allen A.S.
        • Goldstein D.B.
        Genic intolerance to functional variation and the interpretation of personal genomes.
        PLoS Genet. 2013; 9e1003709
        • Zhu X.
        • Need A.C.
        • Petrovski S.
        • Goldstein D.B.
        One gene, many neuropsychiatric disorders: Lessons from Mendelian diseases.
        Nat Neurosci. 2014; 17: 773-781
        • Pergola G.
        • Di Carlo P.
        • D’Ambrosio E.
        • Gelao B.
        • Fazio L.
        • Papalino M.
        • et al.
        DRD2 co-expression network and a related polygenic index predict imaging, behavioral and clinical phenotypes linked to schizophrenia.
        Transl Psychiatry. 2017; 7e1006
      2. Gerring ZF, Gamazon ER, Derks EM, Major Depressive Disorder Working Group of the Psychiatric Genomics Consortium (2019): A gene co-expression network-based analysis of multiple brain tissues reveals novel genes and molecular pathways underlying major depression. PLoS Genet 15:e1008245.

        • GTEx Consortium
        • Ardlie K.G.
        • DeLuca D.S.
        • Segrè A.V.
        • Sullivan T.J.
        • Young T.R.
        • et al.
        Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: Multitissue gene regulation in humans.
        Science. 2015; 348: 648-660
        • Dalmaz C.
        • Barth B.
        • Pokhvisneva I.
        • Wang Z.
        • Patel S.
        • Quillfeldt J.A.
        • et al.
        Prefrontal cortex VAMP1 gene network moderates the effect of the early environment on cognitive flexibility in children.
        Neurobiol Learn Mem. 2021; 185107509
        • de Mendonça Filho E.J.
        • Barth B.
        • Bandeira D.R.
        • de Lima R.M.S.
        • Arcego D.M.
        • Dalmaz C.
        • et al.
        Cognitive development and brain gray matter susceptibility to prenatal adversities: Moderation by the prefrontal cortex brain-derived neurotrophic factor gene co-expression network.
        Front Neurosci. 2021; 15744743
        • Hari Dass S.A.
        • McCracken K.
        • Pokhvisneva I.
        • Chen L.M.
        • Garg E.
        • Nguyen T.T.T.
        • et al.
        A biologically informed polygenic score identifies endophenotypes and clinical conditions associated with the insulin receptor function on specific brain regions.
        EBioMedicine. 2019; 42: 188-202
        • Silveira P.P.
        • Pokhvisneva I.
        • Parent C.
        • Cai S.
        • Rema A.S.S.
        • Broekman B.F.P.
        • et al.
        Cumulative prenatal exposure to adversity reveals associations with a broad range of neurodevelopmental outcomes that are moderated by a novel, biologically informed polygenetic score based on the serotonin transporter solute carrier family C6, member 4 (SLC6A4) gene expression.
        Dev Psychopathol. 2017; 29: 1601-1617
        • Miguel P.M.
        • Pereira L.O.
        • Barth B.
        • de Mendonça Filho E.J.
        • Pokhvisneva I.
        • Nguyen T.T.T.
        • et al.
        Prefrontal cortex dopamine transporter gene network moderates the effect of perinatal hypoxic-ischemic conditions on cognitive flexibility and brain gray matter density in children.
        Biol Psychiatry. 2019; 86: 621-630
        • Restrepo-Lozano J.M.
        • Pokhvisneva I.
        • Wang Z.
        • Patel S.
        • Meaney M.J.
        • Silveira P.P.
        • et al.
        Corticolimbic DCC gene co-expression networks as predictors of impulsivity in children.
        Mol Psychiatry. 2022; 27: 2742-2750
        • McGill M.G.
        • Pokhvisneva I.
        • Clappison A.S.
        • McEwen L.M.
        • Beijers R.
        • Tollenaar M.S.
        • et al.
        Maternal prenatal anxiety and the fetal origins of epigenetic aging.
        Biol Psychiatry. 2022; 91: 303-312
        • Belsky D.W.
        • Domingue B.W.
        • Wedow R.
        • Arseneault L.
        • Boardman J.D.
        • Caspi A.
        • et al.
        Genetic analysis of social-class mobility in five longitudinal studies [published correction appears in Proc Natl Acad Sci U S A. 2018; 115:E10998].
        Proc Natl Acad Sci U S A. 2018; 115: E7275-E7284
        • Lupien S.J.
        • McEwen B.S.
        • Gunnar M.R.
        • Heim C.
        Effects of stress throughout the lifespan on the brain, behaviour and cognition.
        Nat Rev Neurosci. 2009; 10: 434-445
        • Provençal N.
        • Arloth J.
        • Cattaneo A.
        • Anacker C.
        • Cattane N.
        • Wiechmann T.
        • et al.
        Glucocorticoid exposure during hippocampal neurogenesis primes future stress response by inducing changes in DNA methylation.
        Proc Natl Acad Sci U S A. 2020; 117: 23280-23285
        • Folkersen L.
        • Pain O.
        • Ingason A.
        • Werge T.
        • Lewis C.M.
        • Austin J.
        Impute.me: An open-source, non-profit tool for using data from direct-to-consumer genetic testing to calculate and interpret polygenic risk scores.
        Front Genet. 2020; 11: 578
        • Watson J.
        • Jones H.E.
        • Banks J.
        • Whiting P.
        • Salisbury C.
        • Hamilton W.
        Use of multiple inflammatory marker tests in primary care: Using clinical practice research datalink to evaluate accuracy.
        Br J Gen Pract. 2019; 69: e462-e469
        • de Leeuw C.A.
        • Mooij J.M.
        • Heskes T.
        • Posthuma D.
        MAGMA: Generalized gene-set analysis of GWAS data.
        PLoS Comput Biol. 2015; 11e1004219
        • Bulik-Sullivan B.K.
        • Loh P.R.
        • Finucane H.K.
        • Ripke S.
        • Yang J.
        • Schizophrenia Working Group of the Psychiatric Genomics Consortium
        • et al.
        LD Score regression distinguishes confounding from polygenicity in genome-wide association studies.
        Nat Genet. 2015; 47: 291-295
        • Zhu Z.
        • Zhang F.
        • Hu H.
        • Bakshi A.
        • Robinson M.R.
        • Powell J.E.
        • et al.
        Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets.
        Nat Genet. 2016; 48: 481-487
        • Mak T.S.H.
        • Porsch R.M.
        • Choi S.W.
        • Zhou X.
        • Sham P.C.
        Polygenic scores via penalized regression on summary statistics.
        Genet Epidemiol. 2017; 41: 469-480
        • Zheng J.
        • Erzurumluoglu A.M.
        • Elsworth B.L.
        • Kemp J.P.
        • Howe L.
        • Haycock P.C.
        • et al.
        LD Hub: A centralized database and web interface to perform LD score regression that maximizes the potential of summary level GWAS data for SNP heritability and genetic correlation analysis.
        Bioinformatics. 2017; 33: 272-279
        • Hu Y.
        • Lu Q.
        • Powles R.
        • Yao X.
        • Yang C.
        • Fang F.
        • et al.
        Leveraging functional annotations in genetic risk prediction for human complex diseases.
        PLoS Comput Biol. 2017; 13e1005589
        • Robinson M.R.
        • Kleinman A.
        • Graff M.
        • Vinkhuyzen A.A.E.
        • Couper D.
        • Miller M.B.
        • et al.
        Genetic evidence of assortative mating in humans.
        Nat Hum Behav. 2017; 10016
        • Newcombe P.J.
        • Nelson C.P.
        • Samani N.J.
        • Dudbridge F.
        A flexible and parallelizable approach to genome-wide polygenic risk scores.
        Genet Epidemiol. 2019; 43: 730-741
        • Márquez-Luna C.
        • Gazal S.
        • Loh P.R.
        • Kim S.S.
        • Furlotte N.
        • Auton A.
        • et al.
        Incorporating functional priors improves polygenic prediction accuracy in UK Biobank and 23andMe data sets.
        Nat Commun. 2021; 12: 6052
        • Privé F.
        • Arbel J.
        • Vilhjálmsson B.J.
        LDpred2: Better, faster, stronger [published online ahead of print, 2020].
        Bioinformatics. 2020; 36: 5424-5431
        • Liang Y.
        • Pividori M.
        • Manichaikul A.
        • Palmer A.A.
        • Cox N.J.
        • Wheeler H.E.
        • et al.
        Polygenic transcriptome risk scores (PTRS) can improve portability of polygenic risk scores across ancestries.
        Genome Biol. 2022; 23: 23