GenomicSEM Modelling of Diverse Executive Function GWAS Improves Gene Discovery

Area of Science:

• Psychiatric Genetics
• Neuroscience
• Behavioral Genetics

Background:

• Latent variables are the gold standard for measuring executive function (EF).
• Genome-wide association studies (GWAS) typically use singular EF task measures due to logistical constraints.
• Low correlations between individual EF tests raise questions about the construct validity of current GWAS.

Purpose of the Study:

• To investigate the genetic architecture of executive function using factor analysis on GWAS summary statistics.
• To determine if GWAS of EF can capture genetic variance related to latent EF constructs despite using diverse measures.
• To identify novel genomic risk loci associated with common executive function and working memory.

Main Methods:

• Factor analysis was applied to summary statistics from eleven GWAS of EF across five studies.
• The GenomicSEM software was utilized to model the genetic relationships between EF measures.
• Genome-wide association analyses were performed on the derived latent factors.

Main Results:

• A bifactor model was identified, reflecting common EF and working memory-specific variance.
• 20 new genomic risk loci for common EF and 4 for working memory reached genome-wide significance.
• A total of 29 novel EF genes were mapped, exceeding the discoveries of constituent GWAS.

Conclusions:

• GWAS of executive function can effectively measure latent genetic constructs, even with heterogeneous task measures.
• The study clarifies the genetic structure of EF and identifies new genetic associations.
• GenomicSEM enhances statistical power by combining GWAS data from diverse measures of the same phenotype.