Abstract
Human genetics is advancing at an unprecedented pace. Improvements in genotyping technology and rapidly falling costs have accelerated gene discovery. We can now comprehensively scan the genome, testing variation across millions of genetic markers, to identify specific variants associated with any outcome of interest. Large consortia consisting of hundreds of scientists are analyzing data from hundreds of thousands to millions of individuals. Multivariate methods now enable us to identify genes involved in underlying processes, to complement studies focused on specific disorders or traits. There has been an exponential increase in use of direct-to-consumer genetic feedback platforms. These advances are poised to have a widespread effect on medicine and society. However, with such rapid progress will come ethical, social, and legal challenges. Among those challenges are the needs for increased efforts to enhance public understanding of the ways genes contribute to complex behavioral outcomes and for increased diversity in the field of genetics to ensure that all people benefit from advances.
In 2002, as a newly graduated Ph.D., I published a review article on behavior genetics in Current Directions in Psychological Science along with my graduate advisor, Richard Rose. The article was titled “Behavior Genetics: What’s New? What’s Next?” (Dick & Rose, 2002). Twenty years later, when asked to provide an update, I went back to that piece to reflect on how far the field has come. Most of the research directions discussed in that review have stood the test of time: How genetic effects change across the life span (Dick et al., 2006) and how they interact with the environment (Assary et al., 2018) remain active areas of study, as is characterizing the extent to which genetic effects contribute to comorbidity across outcomes (Allegrini et al., 2020). We also forecast advances in gene identification, though it was here that our predictions fell short. We did not foresee the speed and scale at which gene discovery would advance and characterize the recent history of the field. The past decade has been witness to the formation of international consortia consisting of hundreds of scientists scanning millions of genomes to collaboratively advance gene identification. It has been characterized by the launch of the Precision Medicine Initiative (Collins & Varmus, 2015) and rapid escalation in direct-to-consumer genetic testing (Folkersen et al., 2020).
In our review of 20 years ago, Rose and I were also remiss in failing to recognize an insidious problem in the field: the lack of diversity in genetics research—among participants and the research community—and the potential for harm this would create. In this article, I provide an update on advances that have occurred, and, in the spirit of Karl Popper, I make risky new predictions about the potential applications of genetics in the future, which I believe will be far-reaching, with drastic impact on people’s lives. Finally, I discuss the ethical implications that have become increasingly pressing as the field has advanced.
Gene Discovery: From There to Here
Behavior genetics is a field that has experienced rapid advancement the likes of which are unparalleled, with the exception of advances in computer science. The first great triumph of behavior genetics was to create widespread understanding that nearly all behavioral outcomes are under some degree of genetic influence. This represents a drastic shift from the first half of the 20th century, when behavioral disorders were viewed largely as environmentally determined. If the first 50 years of the field were dedicated to demonstrating that a wide range of human behavior is genetically influenced, the next 50 years will surely be consumed with mapping the relevant genes and applying this newfound knowledge.
We have entered a period of unprecedented gene discovery, made possible by advances in genotyping technology, coupled with greater knowledge of genetic variation, and rapidly falling genotyping costs. It is now possible to comprehensively scan the human genome, testing variation across millions of markers, to identify specific variants associated with any outcome of interest. Because scanning millions of genetic markers requires massive correction for multiple testing, it became clear that huge samples sizes would be necessary to conduct well-powered studies. Researchers came together and formed consortia, pooling their samples to achieve huge sample sizes. One of the most prominent examples of these international collaborative consortia is the Psychiatric Genomics Consortium, which consists of more than 800 investigators from more than 40 countries (Watson et al., 2020). Other large-scale consortia have focused on behavioral outcomes, such as alcohol and tobacco use (Liu et al., 2019), and educational attainment (Lee et al., 2018). Several countries and medical systems formed biobanks to collate health information and genotyping for hundreds of thousands of individuals, creating rich resources made broadly available to the field (Bycroft et al., 2018).
The strategy worked. There is compelling evidence that the larger the sample, the more genetic variants are identified (Sullivan & Geschwind, 2019). This has been observed across all psychiatric, substance use, and other complex health conditions and behaviors that have been studied. The sample sizes required to identify genetic variants vary as a function of the heritability of the trait and associated complexities, such as the degree to which the environment plays a role. However, the evidence is unambiguous: Bigger is better. Extremely large, well-powered genome-wide association studies (GWASs) have led to replicable effects, for both individual genes and aggregate genetic liability scores created by summing associated genetic effects across the genome, in a way that earlier candidate-gene, linkage, and smaller association studies did not (Duncan et al., 2019).
In contrast to what Rose and I expected in 2002, phenotyping has taken a back seat in this process. It became clear that having a precise or well-measured phenotype was far less relevant than having a huge sample. However, now that the field has a working strategy for gene discovery, the importance of phenotype is reemerging. There is growing recognition that the genes one finds are substantively affected by the phenotype one analyzes. A salient example can be found in the substance use field, where GWASs for alcohol consumption and alcohol problems produced surprisingly different results, though both alcohol consumption and alcohol problems were considered relevant phenotypes for understanding genetic contributions to alcohol use disorder (Mallard et al., 2021).
Recently, researchers have developed multivariate methods for gene identification (Grotzinger et al., 2019) that allow one to model genetic covariance (the degree of genetic sharing across outcomes) in a way that parallels factor models of phenotypic covariance. These methods allow researchers to use current understanding of the genetic architecture of behavioral traits to aid in gene identification, capitalizing on decades of twin research that elucidates how genetic influences affect behavioral and psychiatric outcomes.
For example, there is compelling evidence that alcohol and other substance use disorders are influenced by a common set of genes that also affect other traits related to self-regulation, such as childhood behavior problems and impulsivity (Krueger et al., 2009). This common underlying factor is called externalizing, and it is highly heritable. On the basis of this literature, my colleagues and I conducted a GWAS of an externalizing factor in 1.5 million individuals, identifying 579 genetic variants that influence self-regulation (Karlsson Linnér et al., 2021). The resulting genetic risk scores, based on how many of the risk variants an individual carries (Bogdan et al., 2018), predict a wide array of behavioral, social, and medical outcomes related to self-regulation in independent samples. They account for approximately 10% of the variance in externalizing behavior, an effect commensurate with the influences of many central social-science constructs, such as socioeconomic status and neighborhood disadvantage.
Science Fiction or Science Fact?
So, what will society do with all this information? Gene identification can lead to enhanced understanding of the underlying biology of disorders, facilitating the development of targeted drug treatments. That process takes time, often decades, before knowledge translates into advances in patient care.
But there is another way GWAS findings can be used, and I predict it will have more immediate impact. The 1997 movie Gattaca (Niccol, 1997) envisions a world where genetic information is mainstream. In an opening scene, a nurse rattles off a baby’s genetic propensity to develop a number of health problems: “Neurological condition: 60% probability; Manic depression: 42% probability; ADHD: 89% probability.” The movie imagines how that information could reshape society and have a profound impact on the lives of individuals who know their genetic proclivities and vulnerabilities from birth.
This represents a drastic shift from the current state. At present, people usually go to their treatment providers when something is wrong, and often they wait to see if the problem resolves on its own, which leads to a worsening of symptoms before they seek help. The idea of precision medicine (Collins & Varmus, 2015), set forth by President Obama in his State of the Union address in 2015, is to convert health care from a treatment-based model to one that is predictive, preventative, and personalized. Because individuals’ DNA sequences are present from birth, genes provide information about outcomes for which a person is at elevated risk, and that information can be used for preventive intervention.
Society is now approaching a time when it is possible to generate risk predictions of the very sort imagined in Gattaca. Presently, there are wide confidence intervals around those predictions, and most genetics researchers will tell you they are not ready for prime time. You will not see genetic risk scores for complex behavioral outcomes in the clinic any time soon. But medical professionals are no longer the keepers of knowledge. Start-up companies have bypassed traditional gatekeepers in many industries, from taxi service to the hotel industry to the medical establishment, to give consumers direct access. The field of genetics is no exception: There has been exponential growth in direct-to-consumer genetic testing (Regalado, 2019) and the use of free public websites that allow users to upload raw genetic data to compute genetic risk scores for a variety of conditions. Half of the top 12 most accessed genetic risk scores from one widely used site are scores for substance use or mental-health outcomes (Folkersen et al., 2020). People want this information, and they can already get it.
If the idea of genetics drastically changing society sounds unrealistic to you, watch a 1994 clip from The Today Show in which the hosts are discussing a mysterious new concept called the Internet (see ‘What is Internet?’, 2019). Finally, a confused Katie Couric asks an off-camera producer, “Can you explain what Internet is?” Fast-forward just 28 years, and most of us carry supercomputers in our pockets. The Internet has changed the way we communicate, shop, and vote. An entire generation of children do not remember a world without limitless information available at their fingertips. Advances in computing have radically changed society, for better and worse, and we are struggling to catch up.
These rapid changes are a product of Moore’s law, which refers to the exponential growth that the computing industry has experienced over the past decades. This “law” is named after Intel cofounder Gordon Moore, who in 1965 observed that computing capacity was doubling every other year. Exponential growth changes our lives and society at a pace that human brains, evolved to think linearly, cannot comprehend. Now consider that genotyping costs have dropped at a pace that surpasses Moore’s law; genetics is just a few decades behind computing with respect to when it entered its exponential growth period. It took $2.7 billion and 13 years to sequence the first human genome, published in 2003; today, it is possible to sequence a human genome for less than $1,000, in a single day. I predict that genetics will be the next major societal revolution. As did advances in computing, this revolution will sneak up on us, and we will find ourselves in a radically changed world for which we are largely unprepared.
DNA Is Not Destiny
So how do we get ready? We are spending billions on gene identification, but comparatively little research is focused on how we will deliver personalized genetic information to individuals in ways that are understandable and empowering. Data indicate that approximately 80% of people want access to their genetic risk scores (Driver, Kuo, Dick, & Spit for Science Working Group, 2020), but more than a quarter of people do not properly understand how complex genetics works. They erroneously believe that genes are determinative and misunderstand the role of the environment (Chapman et al., 2019; Driver, Kuo, & Dick, 2020).
Although our genotypes contain useful information about our risk profile, genes influence behavior in complex ways. There are no genes “for” substance use disorder, depression, anxiety, or other complex behavioral outcomes. Our genetic codes influence the ways our brains are wired, contributing to differences in temperamental traits, the way we respond to risk and reward, our emotionality. These natural tendencies then affect our interactions with the world: the environments we seek out, the way we perceive our environments, the way the world responds to us (Scarr & McCartney, 1983). These processes are called gene-environment correlation, and they create feedback loops that accentuate genetic influence across time. Further, certain environments can constrain or exacerbate the influence of genetic predispositions, a phenomenon known as gene-environment interaction (Dick, 2011).
The complexity of these processes is ignored by GWASs, which agnostically detect any genetic variant associated with an outcome regardless of how that variant exerts an effect. Some associated variants directly influence our biology to alter risk. For example, variants in the ADH and ALDH genes affect an individual’s ability to metabolize alcohol, which protects against developing an alcohol use disorder (Edenberg, 2007). Other genes create cascades of influence via social pathways (Raffington et al., 2020). Adolescents with a proclivity for sensation seeking may seek out bars and risk-taking peers, elevating their risk for substance use disorder. Compounded risk via environmental pathways is likely one of the reasons why heritability estimates from twin studies are higher than the variance accounted for by genetic effects identified in GWASs. Current genetic liability scores are also limited by the methods used to calculate them, which do not take into account gene-gene interactions or fully capture the range of variation that exists across our genomes.
Understanding the pathways by which our genes confer risk, and the limitations inherent in genetic liability scores, highlights why genetic information is not determinative. Gene-outcome associations are correlated with—and can be interrupted by—environmental processes. Mapping genetic pathways, and understanding how associated genes lead to elevated risk, yields information about how to intervene. This information will be critical for precision medicine, so individuals can understand what it means to be at elevated genetic risk, and what they can do to reduce their risk.
But this is not the genetics typically taught in grade school. Lessons cover peas and Punnett squares, monks and Mendel. They focus on single-gene effects, but not about the ways of complex genetics, whereby thousands of genetic variants come together with the environment to influence important life outcomes.
For individuals to understand and interpret genetic scores, they need to understand the complicated pathways by which our genes influence our lives. This requires researchers to do more to get their findings out of scientific journals and into the hands of people who can use this information. I recently wrote a popular-press book called The Child Code (Dick, 2021) in an effort to bring behavior genetic research to mainstream parenting—to help parents understand how their child’s behavior is influenced by the child’s genetic code, to reduce some of the pressure and judgment that often surrounds parenting, and to help parents identify their child’s unique needs. We should be able to harness the wealth of new genetic knowledge to improve lives, but it will require educating the public (and likely many medical professionals) about the complex ways that genes influence behavior.
The Diversity Problem in Genetics
There is another problem in the field of genetics: the profound lack of diversity, both among participants and among the researchers who are driving the research agenda. Nearly 80% of all gene-identification studies are conducted with individuals of European descent (Popejoy & Fullerton, 2016). This is problematic because genetic risk scores are most accurate for individuals whose genetic background matches that of the research participants used to identify relevant genetic variants (Martin et al., 2017). There are many reasons for this (Lewis & Vassos, 2020; Peterson et al., 2019), but central among them is that “background” genetic variation is a product of the ancestral history of the migration of people around the world. When researchers detect associated variants, often they are not picking up on causal variants, but rather are identifying nearby correlated variants, which will necessarily be connected to an individual’s ancestral background. Differences between groups in social processes and gene-environment interaction also likely affect the portability of genetic risk scores.
There are efforts underway to increase diversity in the samples used for genetic studies, but remedying such extreme imbalance will take time and substantial investment. It will also be necessary to increase the diversity of researchers in this field, to bring critical perspectives on working with diverse populations and the unique considerations surrounding conducting genetic research within groups that have been historically and systemically disadvantaged and mistreated by the research community (Davis, 2021).
Genetics will likely make it into the clinic and mainstream before these problems are remedied, and this has concerning potential to further perpetuate health disparities. This should serve as a call to action to all of us with a commitment to antiracism, to educate ourselves about the history and role of genetics in human behavior so we are prepared to address inequities and misinformation.
Conclusions
The field of genetics has entered a period of rapid advancement that promises to revolutionize our society. There is much good that can follow, including more tailored and effective prevention, intervention, and treatment, and a deeper appreciation for our individuality. But there are also complex challenges that will accompany the integration of genetic information into our lives. In our review of 20 years ago, Rose and I ended by addressing ethical, legal, and social issues to be confronted once genes conferring susceptibility to disorders are identified:
How should information about the nature and meaning of susceptibility genes be conveyed to the media, the public, and the courts? How can erroneous beliefs about genetic determinism be dispelled effectively? Such issues will be even more salient once dispositional genes for normal behavioral variation are identified. (pp. 73–74)
That time has come, but little progress has been made toward answering those questions. Those of us who conduct genetics research have been so excited about advances in gene discovery that we have not slowed to address the complex ethical issues that will inevitably follow. My hope is that bringing more widespread attention to these issues will inspire more researchers, from all areas and backgrounds, to enter the exciting, dizzying, and sometimes frightening realm of human behavior genetics, so we can work together collectively to address these pressing questions.
Recommended Reading
Davis, L. K. (2021, March 17). (See References). A call to action to understand the history of racism in the study of genetics and combat the problem of racist ideas in science.
Fullerton, J. M., & Nurnberger, J. I. (2019). Polygenic risk scores in psychiatry: Will they be useful for clinicians? F1000Research, 8(F1000 Faculty Rev), Article 1293. https://doi.org/10.12688/f1000research.18491.1. An overview of polygenic-risk scores in general terms (including a glossary suitable for informed nongeneticists) and discussion of the use and applications of these scores.
Lewis, C. M., & Vassos, E. (2020). (See References). A review of the many possible ways that polygenic-risk scores could be useful in clinical care.
Murray, G. K., Lin, T., Austin, J., McGrath, J. J., Hickie, I. B., & Wray, N. R. (2020). Could polygenic risk scores be useful in psychiatry? A review. JAMA Psychiatry, 78(2), 210–219. https://doi.org/10.1001/jamapsychiatry.2020.3042. An evaluation of the potential use of polygenic-risk scores across multiple contexts in psychiatry.
Raffington, L., Mallard, T., & Harden, K. P. (2020). (See References). An overview of the use of polygenic scores in psychological research, including ethical, legal, and social concerns.
