Genetic path modeling using genomic structural equation modeling
Abstract
The ability to model complex and multivariate relationships between variables and indicators has been the enduring allure of structural equation modeling (SEM). Our tool genomic structural equation model offers the same... [ view full abstract ]
The ability to model complex and multivariate relationships between variables and indicators has been the enduring allure of structural equation modeling (SEM). Our tool genomic structural equation model offers the same ability to model complex and multivariate relationships between traits, based on the summary statistics generated by genome wide association studies (GWAS). We use genomic structural equation modeling to study whether 1. Educational attainment (EA) mediates the relationship between attention deficit hyperactivity disorder (ADHD) and unhealthy or risky behaviors (smoking behaviors, drinking, speeding, eating related behaviors). Our analysis are based on the latest GWAS and our technique is robust for the fact that these GWAS have considerable sample overlap. We further explore whether ADHD liability is a common cause which contribute between the strong genetic relationship between EA and age at death.
To test the causal directions implied in these models we introduce genetic instruments (both individual SNPs, and polygenic instruments based on concurrently modeling 10-75 SNPs). Our results imply the relationship between ADHD and unhealthy or risky behaviors is mediated by EA, suggesting an educational pathway for intervention. We contrast and compare our findings with those based on pairwise and multivariable Mendelian randomization (MR). Results obtained with genomic SEM are consistent with results obtained using genomic SEM, but the ability to account for sample overlap, and the ability to include any number of variables, and thus known confounders, in the model are advantages over MR. Our conditional analysis of the relationship between EA and mortality reveals that this relationship is weakened substantially when both traits are conditioned on ADHD liability, suggesting ADHD liability as a common cause for the two traits.
Authors
-
Michel Nivard
(Department of Biological Psychology, VU University)
-
Andrew Grotzinger
(University of Texas at Austin)
-
Hill Ip
(Department of Biological Psychology, VU University)
-
Elliot Tucker-Drob
(University of Texas at Austin)
-
Dorret I Boomsma
(Vrije Universiteit Amsterdam)
Topic Areas
Cognition: Education, Intelligence, Memory, Attention , Psychopathology (e.g., Internalizing, Externalizing, Psychosis) , Statistical Methods
Session
SY-5A » Genomic Structural Equation Modeling Provides Insights into the Multivariate Genetic Architecture of Complex Traits (10:30 - Friday, 22nd June, Auditorium)
Presentation Files
The presenter has not uploaded any presentation files.