Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Examining genetic effects on reading related traits through polygenic scores in two independent datasets
Poster B116 in Poster Session B, Tuesday, October 24, 3:30 - 5:15 pm CEST, Espace Vieux-Port
Amaia Carrion-Castillo1, Marie Lallier1,2, Manuel Carreiras1,2,3; 1Basque Center on Cognition, Brain and Language, 2Ikerbasque, Basque Foundation for Science, 3University of the Basque Country
Reading relies on several foundational skills, including manipulating speech sounds, rapid and efficient access to lexical representations and orthographic knowledge. However, the extent to which these skills explain reading performance varies across developmental stages and languages. Genetic factors are also known to influence reading and underlying cognitive processes. How the environmental and genetic factors interact to influence reading outcomes is still largely unknown. Polygenic scores (PGS) are individual-level predictors derived from the sum of effect alleles at a single nucleotide polymorphism (SNP), weighted by the regression coefficient describing each SNP’s level of association with a trait of interest. PGSes can also be used to study the genetic relationship between two traits by making predictions across traits. In this study, we use two large datasets of children to explore the extent to which polygenic scores explain variability in reading and related traits. The first is the longitudinal Adolescent Brain Cognitive Development dataset, for which we use a subset of the baseline data (age range 9-10, English) of N=4,080 from European ancestry. The second is the COEDUCA-BCBL dataset, a cross-sectional sample of children (primary grade 2 to secondary grade 2: ages 6.2-16) with a comprehensive assessment of reading-related traits (in Spanish, N>1,200) (Sanchez-Morán et al. 2018), from which we have derived principal components that relate to phonological and orthographic aspects of reading. We first computed PGSes for both datasets using GWAS summary statistics for dyslexia (Doust et al. 2022), reading (Eising et al. 2022), cognitive performance (Lee et al. 2018) and cortical surface area (Smith et al. 2021). Next, we investigated the proportion of variance explained (%adjusted R2) by each of the PGSes in reading in the two target datasets, and in components related to phonological and orthographic processing in the COEDUCA-BCBL dataset through linear mixed-effect models after accounting for covariates (age, sex, grade and genetic PCs). We adjusted for multiple comparisons within datasets using FDR. Our results show that the best PGS predicting reading outcomes is different in these two datasets. The PGS of cognitive performance explains 5.08% of reading accuracy in English (ABCD) and 1.11% in Spanish (COEDUCA-BCBL), while the PGS of dyslexia explains 2.86% for reading efficiency in Spanish and 4.1% in English. In the Spanish database, where a much in depth characterization of reading-related measures in available, reading-related components show a differential pattern of effects across the PGSes. For instance, the PGS of dyslexia explains 1.7% of the variability in the phonological access component and 1.2% of orthographic component related to letter position while no one of the PGSes significantly predicts phoneme awareness accuracy. In addition, phonological short-term memory component is best predicted by the PGS of cognitive performance (2.37%) and variability in the orthographic component is predicted by the PGS of cortical surface area (1.3%). This work provides a validation of the PGS as indices for reading in two general population datasets, and further allows us to disentangle the extent to which reading-related traits have different genetic weight profiles.
Topic Areas: Genetics, Reading