Drug and Alcohol Dependence 220 (2021) 108535

Contents lists available at ScienceDirect

Drug and Alcohol Dependence

journal homepage: www.elsevier.com/locate/drugalcdep

Short communication

Genomic relationships across psychiatric disorders including substance

use disorders
Abdel Abdellaoui 1, Dirk J.A. Smit 1, Wim van den Brink, Damiaan Denys, Karin J.H. Verweij *
Department of Psychiatry, Amsterdam UMC, University of Amsterdam, 1105 AZ, Amsterdam, the Netherlands

A R T I C L E I N F O A B S T R A C T

Keywords: Background: A recent study investigated the genetic associations and latent genetic structure among eight psy­
Substance use disorders chiatric disorders using findings from genome-wide association studies (GWASs). No data from substance use
Genetic correlations disorders were included, while these represent an important category of mental disorders and could influence the
Psychiatric disorders
latent genetic structure. We extended the original paper by recreating the genetic relationship matrix, graph, and
Genetic structure
latent genetic factor structure, including additional data from substance use disorders.
Dual disorder
Methods: We used GWAS summary statistics of 11 psychiatric disorders, including alcohol dependence, nicotine
dependence, and cannabis use disorder. We estimated genetic correlations between all traits in Linkage
Disequilibrium-Score Regression. A graph was created to illustrate the network of genetic correlations. We then
used the genetic relationships to model a latent genetic factor structure.
Results: Alcohol and nicotine dependence showed significant genetic correlations with several other psychiatric
disorders, including ADHD, schizophrenia, and major depression. Cannabis use disorder was only significantly
associated with ADHD. The addition of substance use disorders resulted in some changes in the latent structure of
the factor model when compared to the original model including eight disorders. All substance use disorders
contributed mostly to Factor 3, a heterogeneous factor with also loadings from ADHD, major depression, Autism
Spectrum Disorders, and Tourette Syndrome.
Conclusions: Alcohol and nicotine dependence show widespread genetic correlations with other psychiatric
disorders. Including substance use disorders in the factor analysis results in some changes in the underlying
genetic factor structure. Given the instability of such models, identified structures should be interpreted with
caution.

1. Introduction
Recently, Lee et al. (2019) published a paper in Cell in which they
combined genome-wide association study (GWAS) data from eight
psychiatric disorders - anorexia nervosa (AN),
attention-deficit/hyperactivity disorder (ADHD), autism spectrum dis­
order (ASD), bipolar disorder (BIP), major depression (MD),
obsessive-compulsive disorder (OCD), schizophrenia (SCZ), and Tour­
ette syndrome (TS) – to identify novel pleiotropic genes and a latent
genetic structure underlying these eight disorders. Using exploratory
factor-analyses, and confirmed by hierarchical clustering analyses, they
identified three (correlated) factors explaining approximately half of the
genetic variation in these eight disorders. The first factor consisted
primarily of disorders characterized by compulsive behaviors (AN, OCD,
TS), the second by mood and psychotic disorders (MD, BIP, SCZ), and
the third by early-onset neurodevelopmental disorders (ASD, ADHD, TS)
as well as MD. The authors conclude that their findings have important
implications for psychiatric nosology, i.e., the classification of psychi­
atric disorders.
Unfortunately, the authors did not include any data from substance
abuse or dependence GWASs in their analyses, while substance use
disorders form an important category of mental disorders. Substance use
disorders are highly impairing, causing great harm to the individual,
their family and friends, and to the society as a whole (Whiteford et al.,
2013). There is evidence for substantial phenotypic and genetic corre­
lations between substance use disorders and other psychiatric disorders
(Vink and Schellekens, 2018). Not including substance use disorders in
the analyses is a potential caveat, in particular because the investigated

* Corresponding author.
E-mail address: karin.verweij@amsterdamumc.nl (K.J.H. Verweij).
1
Shared first author.

https://doi.org/10.1016/j.drugalcdep.2021.108535
Received 7 August 2020; Received in revised form 8 December 2020; Accepted 23 December 2020
Available online 19 January 2021
0376-8716/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
A. Abdellaoui et al. Drug and Alcohol Dependence 220 (2021) 108535

traits now lack most of the externalizing disorders. The limited inclusion
of traits may, therefore, have biased the identified latent structure of
disorders.
We investigated whether the genetic latent structure changes when
using a broader set of psychiatric traits, including additional data from
three GWASs on substance use disorders: alcohol, nicotine, and cannabis
dependence. We investigated the genetic relationships between the 11
traits by recreating the genetic relationship matrix and graph, and by
producing a new latent genetic factor structure.
2. Material and methods
2015).
2.2.1. Genetic relationship graph
A graph was plotted with the absolute genetic correlations as
strength parameter for the edges (connections) using R version 3.5.1 (R
Core Team, 2018) and R package igraph (Csardi and Nepusz, 2006).
Layout followed the Fruchterman-Reingold algorithm. Non-significant
weights (p > 0.05 after Bonferroni correction) were removed. Central­
ity of each trait was assessed with eigenvector centrality (Bonacich,
1987), and reflects whether the trait has strong genetic overlap with
traits that themselves have strong overlap with other traits.

2.1. GWAS data 2.3. Factor analysis

To enable direct comparison with the Lee et al. paper, we used data
from the same eight GWAS studies of psychiatric disorders conducted by
the Psychiatric Genomics Consortium: ADHD, AN, ASD, BIP, MD, OCD,
SCZ, TS (Table 1). For MD we only had access to data excluding
23andMe. In addition, we included GWAS data from three GWASs for
substance use disorders: alcohol dependence (AD; DSM-IV-diagnosis;
Walters et al., 2018), cannabis use disorder (CUD, ICD diagnosis;
Demontis et al., 2019a), and nicotine dependence (ND, Fagerström Test
collected among smokers; dividing the sample in: mild (scores 0–3; N =
14,184), moderate (scores 4–6; N = 9206), and severe (scores 7–10; N =
5287); Hancock et al., 2018). In line with the paper from Lee et al. we
only included data based on individuals of European ancestry for the
substance use disorders. Table 1 presents sample sizes and SNP-based
heritabilities for each disorder, and the reference to the original
publications.
As sensitivity analysis we also performed analyses in which we
included data from three GWASs for non-clinical substance use traits:
alcoholic drinks per week (Liu et al., 2019), cigarettes smoked per day
(Liu et al., 2019), and lifetime cannabis use (Pasman et al., 2018).
Sample sizes and SNP-based heritabilities for these traits can be found in
Supplementary Table 1.
2.2. Genetic correlations
We estimated bivariate genetic correlations between all traits using
LD-score (LDSC) regression (Bulik-Sullivan et al., 2015a). The genetic
correlations are based on the estimated slopes from the regressions of
the product of z-scores from two sets of GWAS summary statistics on the
linkage disequilibrium (LD) score and represent the genetic covariation
between two traits based on all polygenic effects captured by the SNPs.
The LD patterns we used were estimated in the European populations
included in the HapMap 3 reference panel (Bulik-Sullivan et al., 2015b;
Finucane et al., 2015). The LDSC regression analyses were performed
using the 1,290,028 genome-wide HapMap SNPs used in the original
LD-score regression studies (Bulik-Sullivan et al., 2015b; Finucane et al.,
We used the genetic correlations to model the joint genetic archi­
tecture of the 11 psychiatric disorders, following the methods by Lee
et al. (2019) to enable direct comparison. First, an exploratory factor
analysis (EFA) was performed with a predefined three-factor extraction
and promax rotation. This determined the factor structure for the sub­
sequent confirmatory factor analysis (CFA), which we performed in
Genomic Structural Equation Modelling (Genomic SEM, Grotzinger
et al., 2019). Three correlated factors were created, including only paths
to traits with path loadings ≥.20 in the EFA. The diagonally weighted
least squares (DWLS) estimator was used. The full genetic relationship
matrix was used as input. Comparative fit index (CFI) and standardized
root-mean-square residual (SRMR) were used as model fit parameters.
Secondly, we repeated the procedure adding a fourth factor, to establish
whether substance use disorders would form a separate factor from the
other disorders.
3. Results
3.1. Genetic correlations
The bivariate genetic correlations between all traits are presented in
Fig. 1A and the exact estimates, standard errors, and p-values in Sup­
plementary Table 2. Of the initial eight disorders, ADHD and MD
showed the highest genetic correlations with the substance use disor­
ders, with ADHD showing significant genetic correlations with all three
substance use disorders (ADHD-AD: rg = .46, p = 3.8×10− 7, ADHD-ND:
rg = .54, p = 3.3×10-13, ADHD-CUD: rg = .35, p = 7.1×10-4) and major
depression only with alcohol and nicotine dependence (MD-AD: rg = .52,
p = 1.3×10-10, MD-ND: rg = .45, p = 9.4×10-12). Both alcohol and
nicotine dependence showed significant genetic correlations with four
psychiatric disorders, whereas cannabis use disorder showed significant
genetic correlations only with ADHD.
3.1.1. Genetic relationship graph
Fig. 1B shows the genetic relationship graph, displaying the

Table 1
Sample sizes and SNP-based heritabilities as computed using LDSC regression.
Psychiatric Disorder Reference Cases Controls Total Effective Sample Size SNP-based h2 (SE)

ADHD Demontis et al. (2019b) 19,099 34,194 53,293 49,017 .26 (.01)
AN Duncan et al. (2017) 3,495 10,982 14,477 10,605 .33 (.05)
ASD Grove et al. (2019) 18,382 27,969 46,351 44,368 .20 (.02)
BIP Stahl et al. (2019) 20,353 31,358 51,710 49,369 .36 (.02)
MD Wray et al. (2018) 59,851 113,154 173,005 156,582 .08 (.01)
OCD Arnold et al. (2018) 2,688 7,037 9,725 7,780 .40 (.06)
SCZ Ripke et al. (2014) 33,640 43,456 77,096 75,846 .26 (.02)
TS Yu et al. (2019) 4,819 9,488 14,307 12,783 .41 (.05)
AD Walters et al. (2018) 11,569 34,999 46,568 34,780 .10 (.02)
CUD Demontis et al. (2019a) 2,387 48,985 51,372 9,104 .16 (.06)
ND Hancock et al. (2018) NA* NA* 28,677 28,677 .12 (.02)

ADHD = attention-deficit/hyperactivity disorder; AN = anorexia nervosa; ASD = autism spectrum disorder; BIP = bipolar disorder; MD = major depression; OCD =
obsessive-compulsive disorder; SCZ = schizophrenia; TS = Tourette syndrome; AD = alcohol dependence; CUD = cannabis use disorder; ND = nicotine dependence.
NA: Not applicable because nicotine dependence was not analyzed with a dichotomous, but an ordinal, variable.

2
A. Abdellaoui et al. Drug and Alcohol Dependence 220 (2021) 108535

Fig. 1. Genetic relationships between eleven psychiatric disorders.

A. SNP-based genetic correlations (rg) between 11 psychiatric disorders (including 3 substance use disorders) using LD-Score regression. The size of the circles
represents the significance of the p-values, whereas the color represents the magnitude of the genetic correlation (darker = stronger). Asterisks indicate statistical
significance after Bonferroni correction for multiple testing (.05/55 = .0009).
B. Graphical representation of the SNP-based genetic correlations between 11 psychiatric disorders. Each vertex (node) represents a disorder. Edge strength was set to
the value of the genetic correlation between the traits. Only edges representing significant genetic correlations after Bonferroni correction are shown. The width and
coloring of the edges represent the strength of the genetic correlation (absolute values of rg). Positive genetic correlations are presented in blue, negative correlations
in orange.
C. Results of the genetic factor analysis. Each latent genetic factor represents shared genetic variance across the psychiatric disorders that load on it. One-headed
arrows connecting the latent genetic factors to the individual disorders represent standardized loadings, which can be interpreted as coefficients from a regres­
sion of the true genetic liability for the disorder on the common factor. Two-headed arrows between the three latent factors represent their correlations. Two-headed
arrows below the individual psychiatric disorders represent residual genetic variance not explained by the latent factor(s). Standardized parameter estimates are
shown with their standard errors in parentheses. In red we show qualitative changes between the results of our genetic factor analyses with 11 psychiatric disorders
and those of the original analysis by Lee et al. (2019) including eight disorders.

significant SNP-based genetic correlations as determined in LD-score
regression. Similar to the plot by Lee et al. (2019), major depression
takes a central place in the graph with many strong genetic correlations
with other traits, now also including alcohol and nicotine dependence.
ADHD also takes in a central place, with additional significant and
strong genetic correlations with all three substance use disorders. The
central positions of major depression and ADHD were reflected in the
fact that they showed the highest eigenvector centrality scores (0.47 and
0.40, respectively).
3.2. Factor analysis
To verify whether our approach matched that reported before, we
replicated the EFA for eight disorders from Lee et al. (2019) without
substantial differences, with the same good model fit (CFI = 0.97, SRMR
= 0.08) and roughly identical path loadings. Adding the substance use
disorders also yielded a good CFI fit index (0.95), and reasonable SRMR
(0.10). The three correlated latent factors together explained 51% of the
genetic variation in the 11 psychiatric disorders. Fig. 1C shows the re­
sults of the CFA as obtained from Genomic SEM, including paths with
loadings ≥0.20 in the EFA. Qualitative changes when compared to the
original model by Lee et al. are shown in red. The factor loadings largely
remained the same, however, some notable changes appeared compared
to the previous analysis. Similar to the results by Lee et al. (2019), Factor
1 consists primarily of disorders characterized by compulsive/­
perfectionistic behaviors (AN, OCD, and TS), but - in contrast to their
results - we now also find a significant and substantial loading of MD.
Our second factor is characterized by psychotic disorders (BIP and SCZ)

3
A. Abdellaoui et al.
and alcohol dependence (with a weak loading), whereas the loading of
major depression is no longer significant. Our third factor is character­
ized by a large number of disorders, including MD, ADHD, ASD, and TS,
and all three substance use disorders. We now found a significant
negative correlation between Factors 1 and 3.
Adding a fourth factor did not converge. To aid convergence, we
forced the removal of additional path loadings by increasing the
threshold from 0.20 to 0.35, which forced a single loading for each
disorder. This model converged but yielded a poor fit (CFI < 0; RMSR >
1), suggesting that a four-factor model is not adequate.
3.3. Sensitivity analyses
As a sensitivity analysis we also performed the genetic factor analysis
with each substance use disorder added separately to the model with the
eight Lee et al. disorders. Model fit indices can be found in Supple­
mentary Table 3 and Supplementary Fig. 1 shows the results of the CFA
as obtained from Genomic SEM. The addition of alcohol dependence led
to similar changes to what we found when adding all three-substance
use disorders, i.e., MD now loaded on Factor 1 and not on Factor 2,
alcohol dependence loaded on factors 2 and 3, and there was a signifi­
cant negative correlation between factors 1 and 3. Adding nicotine
dependence or cannabis use disorder did not lead to any notable changes
to genetic factor structure compared to the Lee et al. model.
We also repeated the main analyses using the Lee et al. data plus data
from three GWASs for non-clinical substance use traits: alcoholic drinks
per week, cigarettes smoked per day, and lifetime cannabis use. Bivar­
iate correlations and the results of the CFA are presented in Supple­
mentary Figure 2 and model fit indices can be found in Supplementary
Table 3. Significant positive genetic correlations are found between
substance use traits and schizophrenia, major depression, bipolar dis­
order, ADHD, and Autism Spectrum Disorder, but not Tourette Syn­
drome, anorexia nervosa, and OCD. Notable changes to the factor
structure compared to that of Lee et al. where: TS now no longer
significantly loads on F1, cigarettes per day loads on F3, and lifetime
cannabis use on F2. Markedly, alcohol drinks per week did not signifi­
cantly load on any factor.
4. Discussion
We extended analyses of a recent study investigating the genetic
associations and underlying genetic factor structure among eight psy­
chiatric disorders, by including three substance use disorders. We show
that alcohol and nicotine dependence are significantly genetically
correlated with several of the other mental health disorders, including
ADHD, schizophrenia, and major depression. Cannabis use disorder is
only significantly associated with ADHD, but this may be due in part to
the limited power of the cannabis use disorder GWAS. The considerable
shared genetic influences of alcohol and nicotine dependence with
psychiatric disease may help explain why their co-occurrence in the
form of dual disorders is significantly higher than would be expected by
chance (Szerman and Martinez-Raga, 2015).
The addition of the substance use disorders to the genetic factor
analysis provides additional information on the psychiatric three-factor
genetic structure identified by Lee et al. (2019). Factor 1 (AN, OCD, TS,
and MD) and 2 (BIP, SCZ, AD) remained quite similar, but MD now loads
on Factor 1, whereas in the original model MD loaded on Factor 2
(although MD still contributes to Factor 2 through the inter-factor cor­
relations). In addition, alcohol dependence now also significantly con­
tributes to Factor 2. Factor 3 is characterized by its large number of
disorders, including all the disorders identified in the original structure
(TS, MD, ADHD, ASD) plus all three substance use disorders. A
four-factor model did not fit the data well, suggesting that substance use
disorders do not form a latent factor separate from the previously
identified factors or that substance use disorders together with ADHD
form a simple externalizing genetic factor. Lastly, we now observe a
4
Drug and Alcohol Dependence 220 (2021) 108535
significant negative correlation between Factor 1 and 3; in the original
model by Lee et al. these factors were also negatively correlated, but the
correlation did not reach significance. This initial negative correlation
was strengthened by adding substance use traits, all of which showed
negative correlations with OCD, and to a lesser degree with AN.
Accordingly, the addition of the substance use disorders to the genetic
factor analysis consolidated and strengthened the results and suggests
that liability to OCD and AN may be protective for the ADHD/substance
use factor. Certain characteristic features of OCD and AN (perfectionism,
inhibition, high educational attainment) may underlie these protective
effects (Frost and Steketee, 1997; Pollack and Forbush, 2013; Yilmaz
et al., 2020; Zucker et al. (2011).
While Lee et al. referred to their third factor as consisting mainly of
early-onset neurodevelopmental disorders (ASD, ADHD, TS), the new
factor 3 is more heterogeneous. It includes four externalizing disorders
(ADHD, AD, ND, CUD) as well as one internalizing disorder (MD) and
other less clearly positioned disorders (ASD and TS). Major depression is
a very common mental health disorder and associated with the broad
range of psychiatric problems, so its contribution to multiple factors,
including this factor, is not particularly surprising. Note that sensitivity
analyses pointed out that changes in the model were due to the addition
of alcohol dependence; adding nicotine dependence or cannabis use
disorder solely did not result in any changes.
The switch of major depression contributing significantly to Factor 2
in Lee et al.’s factor model and to Factor 1 in our model highlights the
sensitivity of such models to the input data. Aside from the additional
input data for substance use traits, the input data for major depression in
our model was a subset of the data used by Lee et al. (i.e., their meta-
analytic sample without 23andMe data), which could also explain
some of the differences in the model. This instability was further high­
lighted when non-clinical substance use traits were added to the Lee
et al. model, resulting in different changes in underlying factor structure.
Accordingly, updating source data (when larger GWASs come out) and
adding additional disorders will likely result in additional changes or
refinements in the underlying genetic factor structure. Larger GWASs
will result in more power to detect significant genetic associations,
changing the genetic correlation matrix and the genetic relationship
graph. For example, the cannabis use disorder GWAS has substantial
genetic correlations with some other traits (Supplementary Table 1), but
these were not significant likely due to the limited sample size of the
GWAS. Furthermore, it is likely that including GWAS data on other
diagnostic data, such as personality disorders, sleep-wake disorders,
panic disorders, sexual dysfunction, neurological disorders, as well as
non-clinical continuous/dimensional data for substance use traits and
mental health problems, would lead to further changes in the underlying
structure. For some of these, and other, psychiatric disorders there are
currently no well-powered GWASs available focused on clinical diag­
nostic information. Once these are available, a more comprehensive and
accurate underlying structure could be identified that may indeed have
important implications for psychiatric nosology. For example, the
identified genetic associations and underlying structure may change the
focus from the distinct clinical diagnoses currently presented in most
diagnostic manuals to more genetically and biologically-informed fluid
disease presentations, with more attention for trans-diagnostic symp­
tomatology and underlying pathways. By further increasing the sample
sizes of GWASs on psychiatric and substance use disorders, we could
potentially identify more stable underlying factors, which could aid in
the detection of novel pleiotropic loci, further elucidate biological
mechanisms that underlie comorbidity, improve traditional nosological
delineations of psychiatric and substance use disorders, and potentially
help explain the high incidence of dual disorders. Until then, given the
instability of the underlying genetic structure, the identified structure by
Lee et al. and those identified in the current study, as well as the inferred
implications for psychiatric nosology, should be interpreted with
caution.
In conclusion, our study illustrated that alcohol and nicotine
A. Abdellaoui et al.
dependence show widespread genetic correlations with other psychiat­
ric disorders. Including substance use disorders in a genetic factor
analysis results in changes in the underlying genetic factor structure of
psychiatric disorders. Our study highlights the sensitivity of such (fac­
tor) models to the input data. Given the instability of such models,
identified structures as well as the implications for psychiatric nosology
should be interpreted with caution. To better estimate the most
comprehensive and accurate underlying genetic structure, future studies
should include data on the broadest possible set of psychiatric disorders.
Role of funding source
Nothing declared.
Contributors
AA, DJAS, and KJHV were responsible for the study concept and the
design of the study. AA and DJAS performed the data analyses, under
supervision of KJHV. KJHV, AA, and DJAS drafted the manuscript.
WvdB and DD provided critical revision of the manuscript for important
intellectual content. All authors approved the final version for
publication.
Declaration of Competing Interest
No conflict declared.
Acknowledgements
A.A. & K.J.H.V. are supported by the Foundation Volksbond Rot­
terdam (the Netherlands). A.A. is also supported by ZonMw grant
849200011 from The Netherlands Organisation for Health Research and
Development.
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.drugalcdep.2021.10
8535.
References
Arnold, P.D., Askland, K.D., Barlassina, C., Bellodi, L., Bienvenu, O., Black, D.,
Camarena, B., 2018. Revealing the complex genetic architecture of obsessive-
compulsive disorder using meta-analysis. Mol. Psychiatry 23 (5), 1181-1181.
Bonacich, P., 1987. Power and centrality: a family of measures. Am. J. Sociol. 92 (5),
1170–1182.
Bulik-Sullivan, B.K., Finucane, H.K., Anttila, V., Gusev, A., Day, F.R., 2015a. An atlas of
genetic correlations across human diseases and traits. Nat. Genet. 47 (11),
1236–1241. https://doi.org/10.1038/ng.3406.
Bulik-Sullivan, B.K., Loh, P.-R., Finucane, H.K., Ripke, S., Yang, J., Patterson, N.,
Neale, B.M., 2015b. LD Score regression distinguishes confounding from
polygenicity in genome-wide association studies. Nat. Genet. 47 (3), 291.
Csardi, G., Nepusz, T., 2006. The igraph software package for complex network research.
Int. J. Complex Syst. 1695 (5), 1–9.
Demontis, D., Rajagopal, V.M., Als, T.D., Grove, J., Pallesen, J., Hjorthoj, C., Borglum, A.
D., 2019a. Genome-wide association study implicates CHRNA2 in cannabis use
disorder. Nat. Neurosci. 22, 1066–1074. https://doi.org/10.1101/237321.
Demontis, D., Walters, R.K., Martin, J., Mattheisen, M., Als, T.D., Agerbo, E., 23andMe
Research Team, 2019b. Discovery of the first genome-wide significant risk loci for
Drug and Alcohol Dependence 220 (2021) 108535
attention deficit/hyperactivity disorder. Nat. Genet. 51 (1), 63–75. https://doi.org/
10.1038/s41588-018-0269-7.
Duncan, L., Yilmaz, Z., Gaspar, H., Walters, R., Goldstein, J., Anttila, V., Bulik, C.M.,
2017. Significant Locus and Metabolic Genetic Correlations Revealed in Genome-
Wide Association Study of Anorexia Nervosa. Am. J. Psychiatry 174 (9), 850–858.
https://doi.org/10.1176/appi.ajp.2017.16121402.
Finucane, H.K., Bulik-Sullivan, B., Gusev, A., Trynka, G., Reshef, Y., Loh, P.-R., Farh, K.,
2015. Partitioning heritability by functional annotation using genome-wide
association summary statistics. Nat. Genet. 47 (11), 1228.
Frost, R.O., Steketee, G., 1997. Perfectionism in obsessive-compulsive disorder patients.
Behav. Res. Ther. 35 (4), 291–296. https://doi.org/10.1016/S0005-7967(96)00108-
8.
Grotzinger, A.D., Rhemtulla, M., de Vlaming, R., Ritchie, S.J., Mallard, T.T., Hill, W.D.,
Tucker-Drob, E.M., 2019. Genomic structural equation modelling provides insights
into the multivariate genetic architecture of complex traits. Nat. Hum. Behav. 3,
513–525. https://doi.org/10.1101/305029.
Grove, J., Ripke, S., Als, T., D, Mattheisen, M., Walters, R.K., Won, H., Anney, R., 2019.
Identification of common genetic risk variants for autism spectrum disorder. Nat.
Genet. 51 (3), 431–444.
Hancock, D.B., Guo, Y., Reginsson, G.W., Gaddis, N.C., Lutz, S.M., Sherva, R., Johnson, E.
O., 2018. Genome-wide association study across European and African American
ancestries identifies a SNP in DNMT3B contributing to nicotine dependence. Mol.
Psychiatry 23 (9), 1911–1919. https://doi.org/10.1038/mp.2017.193.
Lee, P.H., Anttila, V., Won, H., Feng, Y.-C.A., Rosenthal, J., Zhu, Z., Posthuma, D., 2019.
Genomic Relationships, Novel Loci, and Pleiotropic Mechanisms across Eight
Psychiatric Disorders. Cell 179 (7), 1469–1482. https://doi.org/10.1016/j.
cell.2019.11.020.
Liu, M., et al., 2019. Association studies of up to 1.2 million individuals yield new
insights into the genetic etiology of tobacco and alcohol use. Nat. Genet. 51,
237–244.
Pasman, J.A., et al., 2018. GWAS of lifetime cannabis use reveals new risk loci, genetic
overlap with psychiatric traits, and a causal influence of schizophrenia. Nat.
Neurosci. 21, 1161–1170.
Pollack, L.O., Forbush, K.T., 2013. Why do eating disorders and obsessive-compulsive
disorder co-occur? Eat. Behav. 14 (2), 211–215. https://doi.org/10.1016/j.
eatbeh.2013.01.004.
R Core Team, 2018. In: R Foundation for Statistical Computing (Ed.), R: A Language and
Environment for Statistical Computing. Vienna, Austria. doi:http://www.R-project.
org.
Ripke, S., Neale, B.M., Corvin, A., Walters, J.T.R., Farh, K.-H., Holmans, P.A., Psychosis
Endophenotypes International, C., 2014. Biological insights from 108 schizophrenia-
associated genetic loci. Nature 511 (7510), 421–427. https://doi.org/10.1038/
nature13595.
Stahl, E.A., Breen, G., Forstner, A.J., McQuillin, A., Ripke, S., Trubetskoy, V., Gaspar, H.
A., 2019. Genome-wide association study identifies 30 loci associated with bipolar
disorder. Nat. Genet. 51 (5), 793–803.
Szerman, N., Martinez-Raga, J., 2015. Dual disorders: Two different mental disorders?
Adv. Dual Diagn. 8 (2), 61–64. https://doi.org/10.1108/ADD-03-2015-0004.
Vink, J.M., Schellekens, A., 2018. Relating addiction and psychiatric disorders. Science
361 (6409), 1323–1324. https://doi.org/10.1126/science.aav3928.
Walters, R.K., Adams, M.J., Adkins, A.E., Aliev, F., Bacanu, S.-A., Batzler, A., Agrawal, A.,
2018. Trans-ancestral GWAS of alcohol dependence reveals common genetic
underpinnings with psychiatric disorders. Nat. Neurosci. 21, 1656–1669. https://
doi.org/10.1101/257311.
Whiteford, H.A., Degenhardt, L., Rehm, J., Baxter, A.J., Ferrari, A.J., Erskine, H.E.,
Vos, T., 2013. Global burden of disease attributable to mental and substance use
disorders: findings from the Global Burden of Disease Study 2010. Lancet 382
(9904), 1575–1586. https://doi.org/10.1016/s0140-6736(13)61611-6.
Wray, N.R., Ripke, S., Mattheisen, M., Trzaskowski, M., Byrne, E.M., Abdellaoui, A.,
Sullivan, P.F., 2018. Genome-wide association analyses identify 44 risk variants and
refine the genetic architecture of major depression. Nat. Genet. 50 (5), 668–681.
https://doi.org/10.1038/s41588-018-0090-3.
Yilmaz, Z., Halvorsen, M., Bryois, J., Yu, D., Thornton, L.M., Zer, S., et al., 2020.
Examination of the shared genetic basis of anorexia nervosa and
obsessive–compulsive disorder. Mol. Psychiatry 25, 2036–2046. https://doi.org/
10.1038/s41380-018-0115-4.
Yu, D., Sul, J.H., Tsetsos, F., Nawaz, M.S., Huang, A.Y., Zelaya, I., Hirschtritt, M.E., 2019.
Interrogating the genetic determinants of Tourette’s syndrome and other tic
disorders through genome-wide association studies. Am. J. Psychiatry 176 (3),
217–227.
Zucker, R.A., Heitzeg, M.M., Nigg, J.T., 2011. Parsing the undercontrol–disinhibition
pathway to substance use disorders: a multilevel developmental problem. Child Dev.
Perspect. 5, 248–255. https://doi.org/10.1111/j.1750-8606.2011.00172.x.