New Genome Study Adds to the Case for Many Autisms

New Method Fills in the Genetic Picture

A study published in March 2026 in Cell Genomics used long-read whole-genome sequencing to find autism-linked genetic variants that older methods missed. Researchers at the University of California San Diego analyzed 267 genomes from 63 families. Compared with the short-read sequencing that has dominated the field, the long-read approach increased detection of gene-disrupting structural variants by 33 percent and of tandem repeat variants by 38 percent, according to the published findings.

Long-read sequencing reads large stretches of DNA in a single pass, which makes complex rearrangements and repeated segments easier to see. The team reported more than 44,000 structural variants, roughly 60 percent of them not detected by standard sequencing. Jonathan Sebat, the senior author and a professor of psychiatry at the UC San Diego School of Medicine, said the technology could eventually double the share of heritability explained by certain variant types, though he framed that as a hypothesis rather than a result.

The authors were explicit about the limits. The cohort is small, and they wrote that larger studies will be needed before anyone can say how much of autism’s unexplained genetic basis long reads will recover. The work was funded by the National Institute of Mental Health, the National Institute on Drug Abuse, and the National Human Genome Research Institute.

“Long-read technologies are game changers in terms of the diverse functional information we can get from a single genome sequence.” – Jonathan Sebat, Professor of Psychiatry, UC San Diego School of Medicine (2026)

Different Set of Genes But a Familiar Pattern

The UC San Diego paper lands in a sequence of similar findings, and the recurrence is the story. In 2022, a SickKids-led team published the largest autism whole-genome analysis to date in Cell, sequencing more than 11,000 individuals drawn largely from the Autism Speaks MSSNG database. That study identified 134 autism-associated genes and found rare ASD-associated variants in about 14 percent of participants with autism. A version of that finding recirculated through trade and consumer news feeds in mid-2026, which is how a 2022 result reappeared as fresh news.

The throughline across both studies is consistent. Each new method, larger cohort, or sequencing technology adds genes and variant types to the catalog without closing the gap. The SickKids team noted that the genetic cause remains unresolved for roughly 85 percent of autistic individuals. Genetic testing entered standard practice years ago, but it returns an explanatory variant in only about 20 percent of cases, a figure cited by researchers on the 2025 Princeton and Simons Foundation subtyping study.

That is the durable finding underneath the headlines. Autism is highly heritable, with a widely cited meta-analysis of twin studies placing the genetic contribution in the range of 64 to 91 percent, yet the specific genetic architecture is spread across hundreds of genes and many variant types, no one of which accounts for more than a sliver of cases.

Splitting Autism Into Subtypes — in Theory

The clearer trend is conceptual. Research through 2025 and into 2026 has moved away from autism as a single condition and toward biologically distinct subtypes. In July 2025, researchers at Princeton and the Simons Foundation reported four subtypes drawn from more than 5,000 children in the SPARK cohort: Social and Behavioral Challenges, Mixed ASD with Developmental Delay, Moderate Challenges, and Broadly Affected. Each carried distinct developmental, medical, and behavioral traits, and different patterns of genetic variation.

The Autism Science Foundation, in its 2025 research review, called biological subtyping one of the clearest scientific themes of the year, with reproducible subgroups that differ in genetic background and in how behavioral features change over time. The stated promise is precision: stratified diagnosis, and supports matched to subtype rather than to a single label.

For the genomics field, that direction is well established. For ABA, it remains a research-stage idea. No subtype framework currently routes a child to a different behavioral treatment protocol, and the assessment instruments that drive ABA programming, the VB-MAPP and the ABLLS-R among them, do not incorporate genetic data. A genetic result today informs medical surveillance, medication decisions, and genetic counseling for families, not the structure of a behavior plan.

What It Means for Providers

For ABA operators, the near-term effect of the genomics wave is close to zero in the treatment room and larger everywhere around it. Behavioral assessment and programming are unchanged. What shifts is the surrounding conversation: payers, referring physicians, and families increasingly encounter autism described as many conditions with distinct biology, and providers fluent in that framing will field more sophisticated questions about what ABA does and does not address.

The subtype research also feeds a longer-running debate about heterogeneity in outcomes, the same heterogeneity that complicates value-based contracting and outcome measurement in ABA. If the field eventually validates subtypes that predict response to behavioral intervention, the implications for assessment and payment would be direct. That validation does not exist yet. Until it does, the genomics story is one to track, not to act on.

AT A GLANCE

SOURCES & REFERENCES