DATA-DRIVEN MODELLING OF RESTRICTED AND REPETITIVE BEHAVIORS REVEAL TRAIT-SPECIFIC GENETIC SIGNALS IN AUTISM SPECTRUM DISORDER

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by significant behavioral heterogeneity, complicating the identification of genetic associations. To address this, ASD traits are often delineated into three subdomains: restrictive or repetitive behaviors (RRBs), language impairment, and reciprocal social interaction difficulties. While these delineations have supported genome-wide association studies (GWAS), findings remain inconsistent, potentially reflecting may limit genotype–phenotype resolution. This study evaluates whether data-driven latent behavioural modelling improves genotype–phenotype resolution for RRBs.
Behavioral and genomic data were obtained from the Autism Genetic Resource Exchange (AGRE). Exploratory factor analysis identified latent behavioural dimensions, and latent profile analysis characterized participant subgroups. Analyses focused on RRBs due to their established latent structure and heritability, revealing two stable dimensions: insistence on sameness (IS) and repetitive sensorimotor behaviours (RSM). Guided by prior evidence implicating chromosome 8p21.2–8p21.1 in RRBs, a preliminary chromosome 8 single-nucleotide polymorphism (SNP) analyses identified trait-specific genetic signals within loci linked to ASD and neurodevelopment, supporting the utility of empirically derived latent traits.
Extending to whole-genome data, IS and RSM were then used as quantitative phenotypes. SNP-based GWAS and copy number variation analyses assessed common, rare, and structural variation, with emphasis on delineating trait-specific genetic architectures across RRB dimensions. Genome-wide analyses identified partially unique genetic profiles for IS and RSM, possibly reflecting biologically meaningful distinctions. This framework provides a foundation for subsequent genome-wide interrogation of common and structural variation across distinct RRB dimensions, with the aim of delineating trait-specific genetic architectures that may underpin clinically meaningful behavioral subtyping.