Psychiatric Genomics 2025: State of the Art and the Path Forward

I. Introduction

At the dawn of the 21st century, psychiatric genomics, at the confluence of genetics, genomics, neuroscience, and psychiatry, evolved dramatically. It has been instrumental in redefining our understanding of psychiatric conditions, moving from subjective symptom-based diagnostics towards a more biologically informed approach. This manuscript aims to provide a comprehensive overview of the strides made in psychiatric genomics, charting its evolution from the early days of candidate-gene studies to the contemporary era of whole-genome sequencing and genome-wide association studies (GWAS), while also anticipating the next phase of research linking genes to biology through functional genomics and single-cell sequencing.

This chapter summarizes the state-of-the-art in psychiatric genomics as of 2025, highlighting the significant methodological advances, key findings, challenges, and solutions. We describe how integrating genomic data with clinical psychiatry has transformed our understanding of psychiatric disorders. Acknowledging that psychiatric genomics is a vast and complex field, we start with key definitions and an orientation that may be helpful as you navigate this and subsequent articles in this volume.

First, consider the general structure of a eukaryotic cell, as shown in Figure 1. In the cell’s nucleus, there are pairs of chromosomes (which vary in number and type by species), one set of which we have inherited from each of our parents. Each chromosome is essentially a very long unbroken double-helical chain of deoxyribonucleic acid (DNA) molecules with a complementarity that binds adenine nucleotides to thymine nucleotides and cytosine nucleotides to guanine nucleotides. Because of this base-pairing, DNA has an inherent “recipe” for faithful replication. Most of our genome sequence is the same across individuals, but we have key differences too. Most of these individual differences are single nucleotide differences (called single nucleotide polymorphisms, or SNPs), though there are other types of variation as well. The different versions of an SNP are called alleles, and a combination of two alleles possessed by an individual (one inherited from each parent) constitutes their genotype at a given locus. The long stretches of DNA are densely packed and wound around proteins called histones; collectively, this DNA/histone complex is the nucleosome. Histones can modulate the chromatin states and associated gene expression by various chemical modifications such as methylation and acetylation. When a cell is “ready for action,” transcription factors in the nucleus cause the target DNA to unwind from its associated histone and the two DNA strands to unpair from each other, so that new single-stranded ribonucleic acid (RNA) transcripts (complementary to one of the two unwound DNA stretches) can be “written.” These transcripts are then “translated” from the language of nucleic acids to that of amino acids, which are then assembled into polypeptide chains. When those polypeptide chains get long enough (>50 amino acids), the chain can fold, loop, bend, and wind, taking on a secondary structure, which we then call a protein. This pipeline constitutes the bare bones of the “central dogma” of molecular biology. Although this simplification omits a vast array of complex steps in the journey from DNA to protein, it may suffice as an orientation to appreciate the other material in this and subsequent articles in this volume.

Figure 1. Genomic Features of the Eukaryotic Cell.

Now, with this primer under our belt, we summarize the state of the science of psychiatric genomics in 2024. We briefly describe the latest developments in the use of GWASs and genome-sequencing to identify genetic variants that account for heritable variation in a range of psychiatric disorders; this will also be covered in more detail in subsequent chapters focused on specific disorders. Beyond the genetic progress, we will also discuss methodological innovations that are helping researchers move beyond the assessment of a static genome and into the dynamic sphere of genome biology to answer the key question of psychiatric genomics, i.e., “How do genes and environments interact to change biology and make individuals susceptible to expressing symptoms of psychiatric illness?”

Genetics

Psychiatric genomics has advanced markedly since completion of the first draft of the human genome in 2001. The Human Genome Project mapped the entire human genome and laid the foundation for genomic medicine, with the initial discovery of genetic variants associated with mental health disorders and the development of advanced technologies that have enabled the exploration of the genome in its entirety. These advancements have facilitated a deeper understanding of the complex interplay between genetic, environmental, and lifestyle factors in the etiology of psychiatric conditions.

Not only did the completion of the draft of the human genome allow us to understand its sequence and structure, but it also ushered in a wave of methodological developments that have been vital in relating genetic variation to complex phenotypes; i.e., those traits, like psychiatric symptoms and disorders, that are genetically influenced but not caused by a single genetic mutation. From over 100 years of twin studies, we know that all psychiatric disorders studied to date have at least some genetic contributions to their susceptibility. The estimated degree of genetic influence can vary widely, from as low as 30% of total susceptibility for some anxiety disorders to more than 80% for major mental illnesses like schizophrenia (SCZ) and bipolar disorder (BD) ^1,2.

Several decades of genetic association analyses (relating specific alleles to psychiatric phenotypes), particularly GWASs, have begun to explain considerable proportions of the heritable variance in a growing number of mental illnesses. Individual SNPs tend to have very small effects on overall susceptibility for a given disorder, and no SNP is either necessary or sufficient to cause a mental illness. In aggregate, however, large numbers of SNPs can begin to account for major portions of the heritability estimated from twin studies. The most common aggregate index of genetic association with a complex psychiatric trait is the “polygenic risk score,” representing a weighted sum of the combined effects of hundreds to thousands of alleles. For example, the best-performing polygenic risk score from the latest GWAS of SCZ explained 24% of the estimated 80% heritability of SCZ determined from twin studies ³. A similar approach has been applied to other disorders, including BD, major depressive disorder (MDD), attention-deficit hyperactivity disorder (ADHD), autism spectrum disorder (ASD), eating disorders (ED), and substance use disorders (SUD). Further details on each of these achievements are provided in subsequent papers in this volume.

A major achievement in psychiatric genetics over the past two decades has been the discovery of molecular evidence for shared genetic contributions to most psychiatric disorders. In fact, the degree of risk-gene sharing among major psychiatric disorders is far greater than would be expected by chance, which indicates that many psychiatric disorders share an underlying heritable component. Put another way, many genetic variants increase risk for psychiatric illness in a broad sense, rather than simply putting individuals at risk for a particular diagnosis. In fact, for some pairs of disorders, like SCZ and BD, it is difficult to identify any genetic risk factors that have specificity, as virtually all variants found to increase the risk for one of these disorders also increase the risk for the other ⁴. Thus, in a sense, psychiatric genetics research has provided ample evidence refuting the Kraepelinian dichotomy that considered these disorders as entirely separate entities. Molecular genetic evidence would suggest that these diagnoses have more in common at the genetic level than what sets them apart. This has major implications not only for our understanding of these illnesses but also for their nosology and differential diagnosis, as well as their treatment, management, and prevention. This also leaves a major challenge for the field of psychiatric genetics in the coming years: identifying unique genetic contributions to specific mental disorders, and/or using genetic data to help inform a revision to the current nosology.

The subsequent papers in this volume will provide in-depth detail on what is known about the genetic contributions to specific psychiatric diagnoses and dimensions, so we paint here with broad strokes. We introduce methods that have taken us from a point two decades ago, when we had literally zero reliable risk genes for psychiatric disorders, to the present where hundreds of genes are reliably associated with disease susceptibility. Further, large proportions of the variance in who becomes affected with a psychiatric disorder is knowable from genetic screening. We have a long way to go before these methods can be extended to reliable use in clinical practice, but speaking as researchers in this field over this entire duration, the progress has been nothing less than astounding.

Genomics

The field has moved towards a more integrated approach, combining genomic data with epigenetic, transcriptomic, and proteomic insights to unravel the multifaceted nature of psychiatric disorders. The advanced genomic techniques and an enhanced understanding of psychiatric disorders pave the way for groundbreaking discoveries and innovative therapeutic approaches.

One of the most notable developments in recent years has been the emergence of functional genomics studies in the brain. These studies have enabled researchers to move beyond identifying genetic variants to understanding how these variations affect brain function and, consequently, behavior and mental health. Identifying genetic variants that regulate the expression levels of transcripts (expression quantitative trait loci, or eQTLs) in the human brain represents one of the most important achievements.

Another groundbreaking advancement has been the application of single-cell transcriptomics in psychiatry. This technique allows for the examination of gene expression at the level of individual cells, providing unprecedented resolution and insights into the cellular heterogeneity of the brain. The application of single-cell transcriptomics has been particularly transformative in unraveling the complexities of psychiatric disorders, shedding light on the specific cell types and states that contribute to these conditions.

This chapter will also delve into functional studies of disease-associated genetic variants of candidate genes, discussing innovative techniques such as Massive Parallel Reporter Assay (MPRA), Spatial Transcriptomics, Self-Transcribing Active Regulatory Region sequencing (STARR-seq), Cell-Village, and Perturb-Seq. These high-throughput functional genomics tools have revolutionized our ability to study the function of genetic variants, offering new vistas in the understanding and treatment of psychiatric disorders.

In addition to these scientific advancements, this chapter will address the translational use of genomic findings in clinical practice. Integrating genomic data into clinical psychiatry holds immense promise for personalized medicine, offering tailored approaches to treatment based on an individual’s genetic makeup.

II. Functional Genomics Studies in Brain

Functional genomics, with its focus on deciphering the roles, relationships, and actions of genes and their products, stands at the forefront of neuropsychiatric research, particularly in the context of understanding the complexities of brain function and its deviations in psychiatric disorders. Given that the brain is considered the primary organ involved in the pathology of mental illnesses, understanding how genes are regulated and function individually and collectively is the foundation of all mechanistic studies of psychiatric disorders. Quickly evolving technologies have made meaningful functional research possible and numerous informative resources have been developed to facilitate such discoveries.

Goals of Functional Genomics in Understanding Brain Function and Psychiatric Disorders

The primary objectives of functional genomics in studying the brain and psychiatric disorders are manifold, aiming to bridge the gap between genetic information and functional outcomes. A critical goal is to elucidate how specific genetic variants, both common and rare, contribute to the risk of psychiatric disorders by affecting synaptic function, neural circuitry, and brain development ⁵. Functional genomics seeks to understand the molecular and developmental mechanisms through which genes influence brain function, including gene expression regulation, post-transcriptional modifications, and protein interactions ^6–10. With the current advanced sequencing and gene-editing techniques such as CRISPR, gene expression regulation is the most highly studied area to date.

Methodological Advances in Functional Genomics

Recent years have seen remarkable innovations in the methodologies applied to functional genomics studies. Coupled with innovative add-ons or modifications, many kinds of sequencing technologies have been created to quantify various molecules, from DNA to RNA ¹¹. Single-cell RNA sequencing (scRNA-seq) has revolutionized our understanding of the cellular complexity of the brain by allowing the characterization of gene expression patterns at the resolution of individual cells, revealing novel cell types and states implicated in psychiatric diseases ^12,13. Techniques such as CRISPR-Cas9 genome editing have emerged as powerful tools for elucidating gene function in neuronal development and psychiatric disorders, enabling precise manipulation of genomic sequences in model organisms and human cells ^14,15. Additionally, the advent of brain cellular models and organoid technology offers a unique window into human brain development and disease modeling, facilitating the study of the effects of genetic variants in a controlled, yet physiologically relevant, environment ^16,17.

Current Data Resources for Functional Genomics Studies

Advanced technologies have greatly accelerated the production of various data and provided massive, comprehensive data resources. The major data types include psychiatric GWAS results; gene- or protein-expression profiles in brain or brain cells; epigenomic elements in the brain; and genetic variants associated with gene and other molecule expression. Epigenomic elements include DNA methylation, histone modifications, chromatin accessibility, transcription factor binding sites, and chromatin folding and organization. Developmental dynamics and spatial variations are also captured.

The Psychiatric Genomics Consortium (PGC) has aggregated extensive genomic data on a wide range of psychiatric disorders, facilitating large-scale GWAS and meta-analyses that have identified numerous risk loci for psychiatric conditions ⁵. The Genotype-Tissue Expression (GTEx) project, which shares transcriptomic data from dozens of tissues sampled from the same individual donors, has been instrumental in providing insights into tissue-specific gene expression and regulation, including the brain ¹⁸. The BrainSpan Atlas of the Developing Human Brain offers an invaluable repository of gene expression data across different stages of brain development, aiding in the identification of critical windows where genetic variants may exert their effects ¹⁹. These data are summarized in Table 1.

Table 1.

List of Major Brain Functional Genomics Databases

Database Name	Data Type(s)	URL
GWAS Catalog	GWAS	https://www.ebi.ac.uk/gwas/
GWAS Atlas	GWAS	https://atlas.ctglab.nl/
Psychiatric Genomics Consortium (PGC)	GWAS, Psychiatric genetics	https://pgc.unc.edu/for-researchers/download-results/
Gene4Denovo	Human de novo mutations	http://www.genemed.tech/gene4denovo/home
BrainEXP	Gene expression, human brain	http://www.brainexp.org/
BrainEXP-NPD	DEG, brain disorders	http://brainexpnpd.org:8088/BrainEXPNPD/index.html
Allen Brain Atlas	Gene expression, Anatomy, cell-type-specific	http://www.brain-map.org/
Brain RNA-seq	Human and mouse brain, Gene expression, cell-type-specific	https://brainrnaseq.org/
BRAINSPAN	Gene expression (developmental stages)	http://www.brainspan.org/
GTEx (Genotype-Tissue Expression)	eQTL, Gene expression	https://www.gtexportal.org/home/
Gene Expression in Cortical Organoids	Gene Expression, Cortical Organoids	http://solo.bmap.ucla.edu/shiny/GECO/
Human Protein Atlas	RNA and protein expression, in tissues and cells	https://www.proteinatlas.org/
ENCODE (Encyclopedia of DNA Elements)	Gene expression, eQTL, DNA methylation	https://www.encodeproject.org/
EpiMap	Epigenome, tissues, and cells	https://compbio.mit.edu/epimap/
MetaBrain	eQTL, Gene expression (brain-specific)	https://www.metabrain.nl/
Brain xQTL Serve	H3k9Ac, DNAm, and Gene expression QTL	https://mostafavilab.stat.ubc.ca/xqtl/
PsychENCODE	eQTL, brain	http://resource.psychencode.org/
Brain eQTL Almanac (Braineac)	eQTL, Gene expression	http://www.braineac.org/
UK Biobank	GWAS, Gene expression, Imaging data	https://www.ukbiobank.ac.uk/
Roadmap Epigenomics Mapping Consortium	DNA methylation, Histone modification	http://www.roadmapepigenomics.org/

Methodological innovations and expansive data resources have begun to reveal the dynamic landscape of functional genomics in the brain, underscoring the complexity of biological processes. These data are instrumental for unraveling the biology underlying psychiatric disorders through data mining and for generating novel, testable hypotheses for experimental biologists. This wealth of information allows for investigating disease biology at a systemic level rather than focusing on individual genes, offering a more comprehensive understanding of the underlying mechanisms.

III. Single-Cell Transcriptomics in Psychiatry

The advent of single-cell or single-nuclei RNA sequencing (sc/snRNA-seq) has ushered in a new era in the study of psychiatric disorders, offering an unprecedented resolution to explore the complexities of the brain at the level of individual cells. Since scRNA-seq and snRNA-seq are highly similar for the major topics to be discussed in this chapter, we use scRNA-seq to represent both here. This section delves into the basics of scRNA-seq, its applications in psychiatric research, and the challenges faced in its implementation, providing a comprehensive overview of how this cutting-edge technology is reshaping our understanding of psychiatric conditions.

Basics of Single-Cell RNA Sequencing

scRNA-seq is a revolutionary technique that allows for gene expression profiling in individual cells, providing a detailed map of cellular diversity within tissues. Unlike traditional RNA sequencing that offers a bulk average of gene expression across thousands of cells, scRNA-seq uncovers the heterogeneity and intricate molecular processes occurring within single cells ²⁰. This high-resolution approach identifies distinct cell types and states, revealing the cellular composition of tissues and the transcriptional dynamics underpinning biological processes and disease states.

Applications of scRNA-seq in Psychiatry

The application of scRNA-seq in psychiatry has provided novel insights into the cellular and molecular architecture of the brain, shedding light on the pathophysiology of psychiatric disorders. By dissecting the transcriptomic profiles of individual brain cells, researchers have identified specific cell types and molecular pathways implicated in conditions such as SCZ, ASD, and BD ¹³. This fine-grained analysis has revealed the cellular heterogeneity of psychiatric disorders, highlighting the role of specific neuronal and non-neuronal populations in disease mechanisms. Furthermore, scRNA-seq has been instrumental in understanding the neurobiological effects of psychiatric medications, facilitating the discovery of biomarkers for disease diagnosis and treatment response.

Up to 2024, a few psychiatric disorders have been studied for mostly snRNA-seq in postmortem human brains (Table 2). Single-cell RNA-seq experiments were also done on cultured organoids of ASD (with 745,000 cells) ²¹; of Prader-Willi syndrome (PWS) patient arcuate organoids (with 12,505 cells) ²²; cerebral organoids from monozygotic twins discordant for psychosis (with 2,782 cells) ²³; and methamphetamine-treated cerebral organoid (with 20,758 cells) ²⁴.

Table 2.

Current snRNA-seq Studies of Postmortem Brains of Neuropsychiatric Disorders

Author & Year	Disorder	Sample Size	# of cells
Velmeshev D. et al. 2019 29	ASD	15 ASD, 16 HC, 8 epilepsy, 7 HC	104,559 cells, 17 cell types
Grubman A. et al. 2019 30	AD	6 AD, 6 HC	13,214 cells, 6 cell types
Del-Aguila et al. 2019 31	AD	3 AD	26,331 cells, 13 clusters
Mathys H. et al. 2019 32	AD	24 AD, 24 HC	75,060 cells after filtering, 8 major cell types
Nagy et al. 2020 33	MDD	17 MDD, 17 HC	80,000 cells, 26 cell types
Lau et al. 2020 34	AD	12 AD, 9 HC	169,496 nuclei, 6 major cell types
Leng et al. 2021	AD	7 AD, 3 HC	42,528 cells from the Entorhinal cortex and 63,608 cells from the superior frontal gyrus, 7 major cell types
Qian Z et al. 2023 35	Six CNS diseases (AD, PD, Epilepsy, etc)	Total 302	~1 million cells, 8 cell types
Soreq L et al. 2023 36	AD	2 AD and 2 HC	25,000, 10 cell types
Chatzinakos C et al. 2023 37	PTSD, MDD	11 PTSD, 10 MDD, 11 HC, replicates 5 per group	575,000 cells, 8 major cell types
Mathys H. et al. 2023 38	AD	427 AD of different stages	2.3 million, 54 cell types
Emani P.S. et al. 2024 39	SCZ, ASD, BD, AD, MDD/PTSD	77 SCZ, 52 ASD, 34 BD, 33 AD, 10 MDD/PTSD	2.8 million nuclie, 28 cell types
Fröhlich. A.S. et al. 2024 40	SCZ, SAZ, BD, or MDD	87 + 32 individuals, patients and HC	~800,000 nuclei, 21 identified cell types and 7 major cell-type clusters
Xie Y et al. 2024 41	MDD	17 MDD, 17 HC	6014 oligodendrocyte lineage (OL) of 4 stages
Liu A et al. 2024 42	AD	24 AD and 24 HC. Replication: 11 AD and 7 HC	125,939 cells, 8 major cell types

Challenges in scRNA-seq

Despite its transformative potential, scRNA-seq faces several technical and interpretational challenges. Technical hurdles include the complexity of sample preparation, the sensitivity of the technique to detect lowly expressed genes, and the management of technical noise and batch effects ^25,26. The data quality in terms of precision and accuracy for quantification of gene expression is a major issue that has been under-estimated, with the consequence of producing many false findings ²⁷. Some molecule types, such as microRNA, could not be detected well in single-cell assays.

Interpretational challenges arise from integrating scRNA-seq data with spatial, functional, and clinical information to construct a comprehensive understanding of psychiatric disorders. Additionally, the bioinformatic analysis of scRNA-seq data requires sophisticated computational tools and approaches to accurately identify cell types, states, and gene regulatory networks, posing a significant challenge given the vast amount of data generated ²⁸. The definition, nomenclature, and criteria of cell types and cell states remain a problem to settle today.

In summary, single-cell transcriptomics represents a powerful tool in the field of psychiatric genomics, offering unparalleled insights into the cellular basis of psychiatric disorders. The application of scRNA-seq has begun to unravel the complexities of brain function and its dysregulation in psychiatric conditions, promising to pave the way for novel therapeutic targets and personalized medicine. However, overcoming the technical and interpretational challenges associated with scRNA-seq will be crucial for fully leveraging its potential in psychiatric research.

IV. Expression Quantitative Trait Loci (eQTLs) in Psychiatric Research

Psychiatric disorders are undeniably complex and multifaceted, involving numerous genetic variants and their interactions with environmental factors. Expression Quantitative Trait Loci (eQTL) analysis identifies associations between genetic variants and gene expression levels, pinpointing genomic regions responsible for regulating gene expression. This analysis reveals the spatiotemporal specificity of gene expression, which varies across different tissues, organs, and developmental stages. Particularly in neuroscience, brain eQTL studies have become fundamental for unraveling the genetic foundations of psychiatric disorders. They offer insights into the intricate relationship between genetic variations and the phenotypic traits they affect. eQTL analyses facilitate the interpretation of GWAS findings, uncover novel risk genes, and investigate causal links within genetic associations. Moreover, they identify gene-expression regulators as potential experimental targets for validation. Notably, eQTLs typically exhibit larger effect sizes than direct disease associations, enabling their detection in smaller sample sizes. However, ensuring data quality and properly controlling for covariates are essential to yield reliable and reproducible eQTL results.

Mapping eQTLs in the Human Brain

Brain eQTL studies are pivotal for understanding how genetic variations influence brain development, function, and the emergence of psychiatric conditions. Researchers can elucidate the spatiotemporal specificity of gene-expression regulation by mapping eQTLs in various brain regions, cell types, and developmental or aging stages. This spatial and temporal specificity is crucial for comprehending the biological context in which genetic variants operate, offering insights into the cellular and molecular mechanisms underlying psychiatric disorders.

Technological advancements in high-throughput sequencing and bioinformatics have significantly propelled brain eQTL mapping. The first brain eQTL study was published in 2007 ⁴³. A dozen eQTL studies of postmortem, fetal brains, and cultured neural cells, have been published to date (Table 3). Thousands of genes have been associated with eQTLs. However, fetal brain, non-European, and non-cortical samples are still very limited or completely missing. Cell-type-specific eQTL analysis is still in its infancy.

Table 3.

Major brain-related eQTL studies

Author & Year	Sample Source	Sample Size	Ancestry	Life Stage
Myers AJ et al. 200743	Human brain, healthy	193	European	Adult
Heinzen EL et al. 200844	Human brain, Blood, healthy	173	European	Adult
Liu C et al. 2010 45	Human brain, SCZ, BD, Major Depression patients, and healthy controls.	164	European	Adult
Colantuoni C. et al. 201146	Human brain, healthy	269	African, Caucasian	Human lifespan
Ramasamy A. et al. 201447	Human brain, healthy	134	European	Adult
O’Brien HE et al. 2018 48	Human brains	120	European	Second trimester
Wang D. et al. 201849	Human brains, psychiatric patients, and controls	1387	European	Adult
GTEx 202050*	49 tissue types, including brain	114 -205 (vary by brain regions)	European and African American	Adult
Aygun N et al. 202151	Human cultured cell	159
Jerber J et al. 202152	Human cultured cell	215
Bryois J et al. 202253	Human brain, healthy	373	European	Adult
Klein Nd et al. 202354	Human brain	84 to 6523 (vary by brain regions)	European and African American	Adult
Wen C et al. 202455	Human brain	672	European, American, and African American	Fetal

^*based on the latest GTEx v8 summary on February 2024.

Applying eQTLs to Psychiatric Genetics

eQTL analyses have proven instrumental in psychiatric genetics. There are several distinct uses of eQTL findings in psychiatric genetics, including interpretation of GWAS signals, predicting risk genes, solving causal relationships, and suggesting experimental targets.

Explaining GWAS signals: Since more than 80% of GWAS signals are located at non-coding regions of the genome, it is commonly conceived that gene-expression regulation may be the target effects of those GWAS variants. Therefore, eQTLs are the most parsimonious explanation for GWAS signals. If a SNP is associated with disease and gene expression in the brain, the SNP is a strong candidate. However, considering linkage-disequilibrium (LD) relationships among SNPs, fine-mapping methods can capture the most likely SNPs regulating gene expression and disease risk even if the eQTL and GWAS may point to distinct SNPs in a genomic region.

Identifying Novel Risk Genes: Leveraging the relationship between SNPs and gene expression, computational models have been designed to infer gene expression levels from genotypic data. Such models enable the prediction of brain gene expression across large populations, facilitating case-control comparisons to pinpoint genes potentially exhibiting differential expression in affected individuals. This method, known as transcriptome-wide association studies (TWAS), employs tools like PrediXcan ⁵⁶, MetaXcan ⁵⁷, and MR-JTI ⁵⁸ to integrate genotypes with gene expression data. Utilizing eQTL data derived from diverse brain regions, cell types, and developmental stages, TWAS can be applied to various psychiatric disorders. This approach aims to identify risk genes that are specifically active in certain spatiotemporal contexts, revealing candidates that might not be detected through adult-brain eQTL-based TWAS alone. For instance, TWAS focusing on fetal brain eQTLs has successfully uncovered genes overlooked by analyses using adult brain eQTLs.

Despite these advances, the field still faces challenges regarding the limited diversity of available brain eQTL datasets.

Causal Analysis: Distinguishing Association from Causation

Causal inference in observational studies is challenging due to the potential for confounding variables and reverse causation. Mendelian Randomization (MR) emerges as a robust method to ascertain whether an SNP influences disease outcomes through the modulation of gene expression. It capitalizes on the random distribution of alleles at conception, akin to the random allocation in clinical trials, to deduce causal relationships from observational datasets. Various algorithms support MR analyses, including one-sample MR and two-sample MR.

One-sample MR conducts simultaneous analyses of eQTL and GWAS data within the same cohort. This approach inherently minimizes bias related to disparities in study design, population demographics, or measurement techniques since the genetic variants (exposures) and the outcomes are assessed in an identical set of participants. Nonetheless, obtaining datasets encompassing eQTL and GWAS information from the same cohort can be challenging.

In contrast, two-sample MR amalgamates eQTL data and GWAS findings from separate cohorts. This method enhances the analytical power by leveraging more extensive datasets, facilitating the detection of causal links. The PGC provides a typical example of the large GWAS datasets frequently utilized in two-sample MR investigations. An added advantage of the two-sample MR approach is its potential to diminish sampling bias by employing different populations for the exposure and outcome measurements, thereby offering a broader perspective on the causal relationship. Leveraging eQTL data of multiple tissues, a joint-tissue imputation (JTI) Mendelian randomization (MR-JTI) method was introduced to improve TWAS performance in a tissue-sensitive manner.

Experimental Validation: Both association and causal analyses rely on statistical interpretations from population data. It is essential to acknowledge that these findings may be subject to inaccuracies due to overlooked covariates, suboptimal data quality, and lenient statistical methodologies, among other factors. To substantiate the robustness of these findings, the pursuit of independent replication is crucial. However, the gold standard for confirming the validity of such discoveries remains experimental validation under meticulously controlled conditions. Identifying causal variants and their linkage to potential epigenomic regulators, such as transcription-factor binding sites, promoters, or enhancers, highlights genomic areas for further investigation. These regions can then be explored through various experimental approaches, including cellular models, animal studies, or direct biochemical assays, to validate the functional relevance of the genetic findings.

Challenges and Future Directions

Despite the significant promise of eQTL analysis in psychiatric research, several challenges remain. First and foremost is the need for large, well-characterized sample sizes to ensure the robustness and reproducibility of eQTL findings. eQTLs often exhibit larger effect sizes than disease associations, facilitating their detection in relatively small samples. However, the heterogeneity of brain tissue, the complexity of its cellular composition, and the dynamic nature of gene expression across different life stages and environmental contexts necessitate large datasets to capture the full spectrum of eQTL effects. Moreover, controlling for covariates such as age, sex, ancestry, postmortem interval, and cellular composition of brain tissues is crucial for producing reliable and reproducible eQTL results.

Several studies have questioned the effectiveness of using eQTLs to elucidate the findings from GWAS. Small overlaps between eQTLs and GWAS signals were reported. It is recognized that “baseline” or “standard” eQTLs detected under steady-state conditions, typically in adult tissues or cell lines, are likely different from “conditional” eQTLs that can be detected under specific spatiotemporal or environmental conditions. Conditional eQTLs are largely equivalent to what was called “response eQTLs” or “dynamic eQTLs” by Umans et al. ⁵⁹, indicating a significant gene-environment interaction effect.

The regulation of gene expression is complex, involving numerous processes, molecules, and events. eQTL represents merely one small aspect of this big system. Other types of QTLs, such as splicing QTLs (sQTLs), protein QTLs (pQTLs), alternative polyadenylation (apa) QTLs (3’aQTLs)⁶⁰, isoform QTLs (isoQTLs), methylation QTLs (mQTLs), and those related to fetal development, or eQTLs active under specific conditions, remain underexplored. These regulations are anticipated to be governed by genetic factors influencing disease susceptibility. Our investigations into the QTLs of various molecules across multi-omics dimensions ⁶¹ highlighted both the challenges and the excitement of this field. QTLs of the other -omes and conditions may carry substantial coverage of GWAS signals that have been missed by current standard eQTL.

V. Functional Studies of Disease-Associated Variants

Prioritization and Methods

With the identification of numerous genetic variants associated with psychiatric disorders, the next step involves prioritizing these variants for functional studies. This is crucial for understanding their role in disease mechanisms. Techniques for studying genetic variant functions have evolved significantly, incorporating bioinformatics tools and experimental approaches. GWAS, followed by statistical analyses and in silico functional prediction models, have prioritized variants for further study.

Prioritizing candidate genes or SNPs from GWAS involves several advanced methodologies, including fine-mapping, colocalization, and causal analysis methods. These methods enhance our understanding of the genetic basis of complex diseases and traits, informing drug development and personalized medicine strategies.

Fine-mapping techniques, including PAINTOR ⁶², FINEMAP ⁶³, CAVIAR ⁶⁴, and SuSiE ⁶⁵, leverage LD patterns and association signals to narrow down candidate SNPs. For instance, a multi-ancestry GWAS of major depression employed fine-mapping and transcriptome-wide association studies, identifying 205 novel genes significantly associated with the condition ⁶⁶. SuSiE-inf and FINEMAP-inf ⁶⁷ have been introduced to improve the reproducibility and functional enrichment of identified causal variants. The PICS2 algorithm is featured for fine-mapping causal SNPs on index SNP-only data, where full summary statistics or genotypes are not available ⁶⁸.

Colocalization analysis compares the disease GWAS signals with eQTL to assess whether they share a common genetic cause. This method helps distinguish pleiotropic effects (a single gene influencing multiple traits) from coincidental associations within the same locus. COLOC ⁶⁹, HyPrColoc ⁷⁰, enloc ⁷¹, Ecaviar ⁷², and eQTpLot ⁷³ are a few commonly used colocalization methods. These methods have successfully identified potential causal genes for SCZ ^74,75, ASD ⁷⁶, and BD ⁷⁷. Causal-TWAS (cTWAS) was developed ⁷⁸ recently to address the genetic confounders, reducing false findings.

High-Throughput Techniques

With large numbers of candidate variants identified, functional experiments now need high-throughput techniques. MPRA allow the assessment of gene expression regulation activity of thousands of DNA sequences simultaneously ⁷⁹. Similarly, STARR-seq has been utilized to discover DNA sequences with gene expression-regulation activity ⁸⁰. Innovative approaches like Cell-Village ⁸¹ and Perturb-Seq ⁸², Mosaic-seq ⁸³, CRISP-seq ⁸⁴, and CROP-seq ⁸⁵ combine single-cell RNA sequencing with CRISPR-Cas9 mediated genome editing to study the effects of genetic perturbations on gene expression in a high-throughput format. However, these techniques face certain challenges too. For example, all of them use specific cultured cells, with inherited limitations of cellular models. Moreover, not all cell types and tissues have good cellular models yet.

Functional Studies of Candidate Genes

Classical cellular models (2D), 3D organoid models, and biochemical experiments are the most commonly used methods to study functions of candidate genes. Functional studies of selected genes have begun to reveal how specific genetic variants contribute to psychiatric disorders (Table 4).

Table 4.

Psychiatry GWAS or CNV candidate genes that have been studied by cellular and animal models

Authors	Gene	Models (2D/3D)
Hill MJ et al.86	ZNF804A	2D
Zeng L et al.87	NRXN1	2D
Srikanth P et al.88 Wen Z et al.89	DISC1	2D
Duan J et al.90	Mir-137 /mir2682	2D
Won H et al.91	FOXG1	3D
Hill MJ et al.92	TCF4	2D
Meng Q et al93	DGCR5	2D
Yang CP et al.94	GLT8D1 and CSNK2B	2D
Chen C et al.95 Ding C et al.96	POU3F2	2D
Cameron D et al.97	SETD1A	2D
Schrode N et al.98	FURIN, TSNARE1, CLCN3, SNAP91	2D
Meng Q et al.99	GNL3	2D and 3D
Ding C et al.96	TRIM8	2D
Reviewed by Zhang C et al.100	ANK3 and CACNA1C	mouse, human, 2D
Wang L et al.101	NLGN3	Mouse and 2D
Kozlova A et al.102	OTUD7A	2D
Zhang W et al.103	PCCB	2D and 3D
Zhang S. et al.104	VPS45	2D

Multiple other neuropsychiatry-related genes impacted by copy-number variations, including NRXN1, CYFIP1, PCDH15, RELN1, and SHANK3, have been reviewed by Hoffmann et al¹⁰⁵

Challenges and Opportunities

Despite the advancements, functional studies of disease-associated variants face several challenges. One major challenge is the complexity of psychiatric disorders, which are influenced by numerous genetic and environmental factors. Of the numerous factors, each has small effects on disease risk. Environmental factors in psychiatric disorders have been studied only rarely in cellular models thus far. Moreover, molecular and cellular functional assays can be resource-intensive and difficult to scale. However, these challenges also present opportunities for innovation. The challenges demand novel technologies with high throughput, resolution, sensitivity, accuracy, and precision. High throughput enables the collection and study of large samples, which is essential to uncovering weak-effect risk factors. High resolution is critical for resolving spatiotemporal complexity. High sensitivity is needed for studying low-abundance factors. High accuracy and precision enhance the statistical power, making discovery possible when sample size is limited. New statistical methods, machine learning models, and algorithms are needed to process and analyze the increasingly complex multi-dimensional data.

VI. Emerging Trends and Future Technologies

Biomarkers and Diagnostic Tools

Objective biomarkers are essential for revolutionizing psychiatry. The search for reliable biomarkers and the development of innovative diagnostic tools are at the forefront of psychiatric genomics.

Advances in sequencing technologies and functional genomics are enabling the discovery of biomarkers that can provide insights into the biological underpinnings of psychiatric disorders. These biomarkers have the potential to develop diagnostic tests that can offer early and more accurate diagnoses, guide treatment decisions, and monitor response to therapy. The search for peripheral biomarkers has yielded many leads, but even the best multi-biomarker classification profiles are not yet sensitive or specific enough for clinical use in diagnosing mental disorders or monitoring their treatment.

Epigenetics

Epigenetics is another under-explored field of study, but it has a high potential for biomarkers. Epigenetic modifications, which regulate gene expression without altering the DNA sequence, are increasingly recognized for their role in psychiatric disorders. Epigenetic modifications are products of gene-environment interaction. They can be dynamically responsive, reflecting state changes associated with disease progress and treatment responses. However, epigenetic regulation in the brain remains poorly studied in psychiatry.

AI and Machine Learning

Integrating artificial intelligence (AI) and machine learning (ML) into psychiatric genomics represents one of the most transformative trends in the field¹⁰⁶. AI and ML algorithms have the potential to parse through genetic, epigenetic, transcriptomic, and phenotypic data to uncover novel risk factors and their regulatory frameworks; predict disease susceptibility; and identify potential therapeutic targets. By identifying complex non-linear patterns undetectable to the human eye, or even advanced statistical modes, these technologies may improve the accuracy of prediction models for diagnosing psychiatric disorders and enabling more personalized approaches to prevention and treatment.

VII. The Path Forward

Challenges of Complexity, Causation, and Solutions

The advancement of psychiatric genomics faces numerous challenges that must be addressed to realize its full potential. One of the primary barriers is the complexity of psychiatric disorders themselves, which are influenced by a vast array of genetic, environmental, and lifestyle factors. This complexity necessitates the development of sophisticated analytical tools and methodologies capable of deciphering the intricate interactions that contribute to these conditions. The environmental factors of psychiatric disorders have been difficult to track and measure. Quantitative study of genomic response to environmental challenges is lacking. Additionally, there is a critical need for large-scale, diverse genomic datasets to enhance the generalizability of findings across populations. Addressing these challenges requires not only advancements in technology but also a concerted effort to ensure that study populations are representative of global diversity, thereby reducing health disparities. For example, including a diverse population enhances the power of discovering novel risk genes or regulatory elements ¹⁰⁷.

Another significant challenge is to turn association to causation to guide translational studies. While numerous genetic variants associated with psychiatric disorders have been identified, the functional relevance of many of these findings remains unclear. Enhancing our understanding of the biological mechanisms underlying these associations is crucial for developing targeted therapies and interventions, laboratory-based diagnostics, and early identification protocols. This necessitates a multidisciplinary approach, integrating insights from genetics, neuroscience, psychology, and computational biology.

Solutions through Collaboration

The complexity of psychiatric genomics research underscores the importance of collaborative efforts and funding. Large-scale international consortia, such as the PGC, have played a pivotal role in advancing the field by pooling data and resources to GWASs on an unprecedented scale. These collaborative efforts must be sustained and expanded, with adequate funding from the governmental and private sectors, to facilitate sharing of data, tools, and methodologies across research groups. PGC is growing its footprint in functional genomics after many years of investment in increasing sample size for GWAS.

Interdisciplinary collaborations are essential for breakthroughs in psychiatry research. Such collaborations will accelerate the development of new experimental methods and computational models. Building trust and ensuring open access to data and resources while protecting participant privacy and ethical considerations is critical for facilitating these collaborative efforts and advancing the field.

Vision for the next 10 years

Looking toward the future, the vision for future psychiatric genomics is one of optimism and ambition. Predictions for the next decade include a deeper understanding of the genetic basis of psychiatric disorders, enabled by advances in sequencing technologies, functional genomics, and AI-driven analyses. These advancements are expected to lead to the identification of comprehensive risk predictors and objective biomarkers for early diagnosis, subtyping patients, monitoring disease progress, development of new therapeutic targets, and the personalization of treatment strategies based on an individual’s genetic makeup and environmental challenges.

A key goal for the next 10 years is integrating psychiatric genomic research to uncover mechanisms of psychiatric disorders, translating basic research findings towards precision psychiatry. This approach will leverage genomic data to inform the prevention, diagnosis, and treatment of psychiatric disorders, tailoring interventions to each individual’s unique genetic profile.

The future of psychiatric genomics lies in its ability to reduce health disparities and ensure equitable access to genomic medicine. This entails a commitment to including diverse populations in genomic research, developing effective interventions across different genetic backgrounds, and ensuring that the benefits of genomic medicine are accessible to all, regardless of socioeconomic status or geographic location.

Concluding Remarks

The path forward for psychiatric genomics is filled with both challenges and opportunities. Overcoming these challenges will require innovative scientific approaches, collaborative efforts across disciplines, and sustained funding support. Genomic insights will lead to a deeper understanding of psychiatric disorders and herald a new era of precision psychiatry. By addressing the current barriers and leveraging the power of collaboration, the field of psychiatric genomics is poised to make significant strides in the coming decade, improving the lives of individuals affected by psychiatric disorders worldwide.

Key Points:

1. Genetic studies have identified hundreds of genetic variants associated with mental illnesses, with a significant overlap in risk genes among disorders. This challenges traditional classifications and also calls for functional interpretation of causal risk variants.

2. Mapping of expression quantitative trait loci (eQTL) in brain and neural cells has revealed functional elements that regulate gene expression in the human genome. Some of these regulations influence psychiatric pathology.

3. Single-cell and other new technologies and cellular models have revolutionized functional studies of genetic variants and their associated genes in the human brain.

4. Major databases and websites of functional genomics are cataloged here for reference.

5. The complexity of psychiatric disorders calls for data from large cohorts and diverse populations, interdisciplinary and international collaboration, and novel technologies featured with improved throughput, sensitivity, resolution, and advanced analytical tools. The primary goals are objective biomarkers, early diagnosis, and the development of targeted therapies.

Synopsis:

Psychiatric genetics has evolved from candidate-gene studies to whole-genome sequencing efforts. With hundreds of disease-associated loci now identified, functional interpretation of the associated loci becomes the critical next step toward translational applications. The chapter discuss achievements, challenges, and opportunities in psychiatric genomics associated with complexity and heterogeneity. Brain eQTL, single-cell RNA-seq, and functional genomics technologies are highlighted. It also covers newly developed techniques with improved spatiotemporal resolution, quality and sensitivity, coupled with advanced analytical methods and artificial intelligence. The power of collaborative research and inclusion of diverse populations will ensure a bright future for precision psychiatry.

Box 1. Definitions.

Gene: the basic unit of heredity stored in cells as a stretch of DNA that encodes a functional element; for example, a gene may define an area of the DNA that is transcribed and then translated into a protein (a coding gene), or an area that is transcribed but not translated (creating a non-coding RNA)

Genetics: the study of DNA sequence variation and its association with traits

Phenotype: a trait whose variability in the population is commonly influenced by genetic factor(s)

Quantitative Trait Locus (QTL): A DNA sequence variation related to a quantifiable (rather than binary) trait. An expression QTL (eQTL) is DNA sequence variation that facilitates or inhibits the transcription of a gene, where the amount and type of gene product expressed is the quantitative trait of interest.

Genome: the full complement of DNA base sequences present in the cells of an organism

Genomics: the study of the structure and function of the genome

Epigenetics: the study of chemical modifications to the genome not involving changes in its DNA base sequence; for example, methylation of cytosine DNA bases, or acetylation of chromosomal histones

Epigenomics: the study of the structure and function of the epigenome

Functional genomics: the study of the functions of genes and their transcribed products

Transcriptomics: the study of all RNA transcripts expressed in a cell or tissue

Proteomics: the study of all translated proteins expressed in a cell or tissue