SHAReD – SFARI Heterogeneity in Autism — Reanalysis of data

Heterogeneity in autism is well documented and complicates efforts to identify biological and social mechanisms underlying the condition. The Varun Warrier lab’s prior work using SFARI and other large-scale resources has demonstrated that autism heterogeneity can be genetically dissociable across domains, stratified by latent diagnostic factors, and influenced by co-occurring intellectual disability (ID) and sex. More recently, Warrier and colleagues have identified age at diagnosis as a key axis of heterogeneity, with early- and later-diagnosed autism showing low genetic correlation (rg ~ 0.3) and distinct associations with ADHD and mental health conditions. These findings suggest that autism polygenicity can be decomposed into partially distinct subtypes, yet critical gaps remain across ancestries, phenotypes and biology.

This proposal addresses three aims. Aim 1 will quantify genetic, phenotypic and biological overlap between early- and later-diagnosed autism across diverse ancestries, leveraging improved statistical power through novel mixed-model approaches, multi-ancestry GWAS, and integration with functional genomics to identify implicated genes, pathways, and cell types. Aim 2 will dissect the polygenic architecture of autism with and without ID/developmental delays (DD), testing competing hypotheses of distinct versus additive aetiology and evaluating convergence between rare and common variant signals. Aim 3 will examine whether core autism features (social-communication and repetitive behaviours) and their genetic influences differ across these axes using factorial invariance and variant–strata interaction models.

By integrating genetic, phenotypic, and functional data across the Simons Simplex Collection, SPARK, the Autism Inpatient Collection and the Genomics England cohort, this work will clarify the structure of autism heterogeneity, establish robust axes of stratification, and test their impact on etiology, biology and core features. Identifying etiologically supported subtypes has direct clinical implications for diagnosis, taxonomy, and support for autistic individuals and their families.