The Multivariate Genome-wide Architecture of Interrelated Literacy, Language and Working Memory Skills Reveals Distinct Etiologies

Abstract

There is genetic overlap between many measures of literacy, language and phonological working memory (PWM) though our knowledge of multivariate genetic architectures is incomplete. Here, we directly modeled genetic trait interrelationships in unrelated UK youth (8-13 years, N=6,453), as captured by genome-wide relationship matrices, using novel structural equation modeling techniques. We identified, besides shared genetic factors across different domains (explaining 91-97% genetic variance in literacy-related measures such as passage reading fluency, spelling, phonemic awareness, 44% in oral language and 53% in PWM), evidence for distinct cognitive abilities; trait-specific genetic influences ranged between 47% for PWM to 56% for oral language. Among reading fluency measures (non-word, word and passage reading), single-word reading was genetically most diverse. Multivariate genetic and residual covariance patterns showed concordant effect directionality, except for near-zero residual correlations between oral language and literacy-related abilities. These findings suggest differences in etiological mechanisms and trait modifiability even among genetically highly correlated skills.