Polygenic risk score
Autism Spectrum Disorder (ASD)
ICD-11 6A02 · Neurodevelopmental
The Autism Spectrum Disorder polygenic risk score summarizes common-variant liability from the 2019 PGC Grove GWAS. Autism is notable for a large rare-variant contribution (de novo mutations, CNVs) that a common-variant PRS cannot capture.
The autism common-variant PRS captures only part of the genetic architecture — rare de novo mutations and large CNVs account for a substantial fraction of individual autism risk, invisible to this score.
At a glance
Autism spectrum disorder is highly heritable, with both rare high-impact variants and many common small-effect variants contributing. Your common-variant score reflects only the latter.
Autism Spectrum Disorder is a neurodevelopmental condition characterized by differences in social communication and restricted/repetitive behaviors. Twin heritability is around 80%. The 2019 PGC-ASD GWAS identified the first robust common-variant signals; rare de novo variants and copy-number changes contribute substantially to ASD risk and are *not* captured by common-variant PRS.
What the GWAS actually found
The Grove et al. 2019 ASD GWAS (Nature Genetics) combined 18,381 autism cases and 27,969 controls, identifying 5 independent genome-wide significant loci. SNP-based heritability was estimated at approximately 11.8% (liability scale). The comparatively modest loci count relative to MDD or schizophrenia reflects both a smaller case cohort and the heterogeneity of the autism spectrum.
Twin-study heritability for autism is high (~0.70–0.90), but a large fraction of this is driven by rare variants — de novo single-nucleotide variants in genes like CHD8 and SCN2A, and large CNVs such as 16p11.2 deletions — that contribute to individual risk far more than any common-variant PRS.
Full citation: Grove J, et al. (2019). Identification of common genetic risk variants for autism spectrum disorder. Nature Genetics 51:431-444. PMID: 30804558..
How to read your percentile
Below 25th
Lower common-variant liability than most of the reference cohort. Does not rule out autism — rare variants and CNVs are independent risk factors invisible to this score.
25th–75th
Typical range — near the population average. PRS is not clinically informative for autism at this percentile.
Above 75th
Elevated common-variant liability. Interpret as one input alongside developmental history, and consider that rare-variant testing (exome sequencing, CNV analysis) is clinically distinct from a PRS.
What this does not tell you
A common-variant PRS cannot capture de novo mutations or CNVs, which together account for an estimated 10–30% of individual autism risk depending on cohort. A low PRS does not rule out autism caused by rare or de novo variation, and a high PRS is not a diagnostic signal.
The score does not distinguish autism subtypes (high-need vs minimal-support, syndromic vs non-syndromic, male vs female phenotype). Autism in females is historically underdiagnosed and may have partially distinct genetic architecture. Cross-ancestry accuracy is reduced outside European populations.
Related traits
Autism Spectrum Disorder (ASD) shares common-variant architecture with several other psychiatric conditions. Genetic correlations (rg) reflect how often the same variants move risk for both traits in the same direction.
Frequently asked questions
Can this score identify autism?
No. A PRS is a population-level probabilistic signal, not a diagnostic tool. Autism diagnosis requires clinical evaluation based on observed behavior, developmental history, and functional assessment — not genetics.
Why does the autism PRS have so few loci compared to schizophrenia?
Two reasons: a smaller case cohort (~18k vs ~76k) and substantial rare-variant contribution to autism risk that common-variant GWAS cannot detect. Larger cohorts and rare-variant sequencing studies will refine autism genetic architecture over time.
Does a low PRS rule out autism?
No. Autism caused by a de novo mutation or CNV can occur in individuals with low PRS. A common-variant PRS misses an estimated 10–30% of genetic architecture depending on the cohort and methodology.
Is there a 'female autism' PRS?
Not yet. Autism GWAS historically include more male cases, and female autism may have partially distinct genetic contributors. Sex-stratified autism GWAS are an active area of research but are not what this report scores.
Sources
- Grove et al. 2019, Nature Genetics 51:431–444. doi.org/10.1038/s41588-019-0344-8
- PGC Autism Spectrum Disorder Working Group. pgc.unc.edu/for-researchers/working-groups/autism-spectrum-disorder-workgroup/
Want your own polygenic risk score?
Upload your AncestryDNA or 23andMe raw data to get the ADHD and Anxiety preview free, or unlock all 13 traits for $29 (launch pricing).
Upload your file