Researchers now count 253 genes strongly linked to autism, up from 72 in 2022. That is real progress for research, but it is not "the cause" of autism, not a test, and not the promise of a treatment. Two pieces of work that made the news in October 2026 help put these numbers in perspective.
- The number: 253 genes found in the DNA of 62,429 autistic people, in a preprint that has not yet been peer reviewed.
- The nuance: in a companion preprint covering 38,680 autistic people, about 7% carry, according to a statistical model, a new (de novo) large-effect mutation in a gene (likelihood of autism multiplied by more than five). The great majority do not.
- The other study: anxiety, depression or ADHD in autistic people seem to be linked to different genetic factors from those behind autism.
- For you: no genetic test can diagnose autism.
From 72 to 253 genes: what the study found
The main work was posted on 26 August 2026 on medRxiv, a preprint server, by the Autism Sequencing Consortium (senior authors: Mark Daly, Broad Institute, and Michael Talkowski, Massachusetts General Hospital). The science magazine The Transmitter covered it on 8 October.
The team read the exome, the part of DNA used to make proteins, of 62,429 autistic people from research cohorts (including SPARK) and a clinical genetics laboratory. They looked for rare variants that damage a gene: some appeared for the first time in the child (de novo variants), others were inherited from a parent.
The result: 253 genes associated with autism at a very strict threshold (fewer than one false discovery expected in 1,000). In 2022, the same team found 72 in about 20,000 autistic people, and 66 of them are confirmed. Another large 2022 study, of 42,607 autistic people, counted 60 with a different method. Three times more participants, more than three times more genes: discovery has not reached a ceiling.
A word on vocabulary: an "autism gene" is not a gene that makes someone autistic. It is a gene in which a rare change clearly raises the likelihood of being autistic, without making it certain.
Genes that are not only about autism
The authors compared these genes with large genetic studies of developmental disorders, schizophrenia, bipolar disorder and epilepsy. They sorted them into six groups: for example, 39 genes are linked only to autism, and 128 to both autism and developmental disorders.
Genes more specific to autism are more often involved in building synapses, the contact points between neurons. Shared genes are more often involved in regulating the activity of other genes. Children carrying a variant in these shared genes reached developmental milestones, such as walking or first words, later.
These are statistical trends: as Mark Daly told The Transmitter, we cannot yet predict how a person's life will unfold from their genetic result.
A companion preprint (bioRxiv, 23 September 2026, Michael Gandal's team) shows that these genes converge on three broad programmes: gene regulation, neuron growth and communication between neurons. They are mostly active in neurons that are still maturing.
Funding: including the Simons Foundation (which also funds SPARK and The Transmitter) and the US National Institutes of Health. Mark Daly is a scientific founder of the company Maze Therapeutics, and several authors are or were employees of GeneDx, the clinical genetics laboratory that provided part of the data.
How many autistic people carry one of these variants?
A third preprint partly answers this (Ajay Nadig, Luke O'Connor and colleagues, medRxiv, 27 August 2026). It models de novo variants in 38,680 autistic people whose parents were also sequenced.
- According to the model, about 7% of them (confidence interval 5.6% to 8.4%) carry a large-effect de novo mutation: a variant that truncates a protein, or a missense variant rated as highly damaging, in a gene where it multiplies the likelihood of being autistic by more than five.
- Taken together, damaging de novo coding mutations, large-effect or not, explain 3.4% of the variation in autism in these cohorts, a group-level statistical measure.
- Half of that share comes from just 15 genes, mostly linked to severe developmental disorders.
- Even these mutations do not always lead to autism: according to the model's estimate, a typical mutation that truncates an autism gene would lead to autism in only about 56% of carriers (estimated range 38% to 68%).
The 7% figure applies to these cohorts, where more than a quarter of participants came from a clinical genetics laboratory, in which intellectual disability and developmental delay are more common: it does not apply as such to all autistic people. This calculation also leaves out large deletions or duplications of DNA, inherited rare variants and non-coding DNA. The share of autistic people with a rare-variant explanation is therefore higher, but still a minority: as researcher Maria Chahrour, who was not involved, pointed out, most autistic people have no rare-variant explanation.
So where does the rest come from?
Autism is highly heritable: about 80%, according to a study of more than 2 million children born in five countries (Bai and colleagues, 2019). This means that, within a population, differences between people are largely linked to genetic differences. It does not mean that one person's autism is "80% genetic".
Most of it comes from common variants: thousands of small DNA differences, carried by many people, each with a tiny effect. A 2014 Swedish study (Gaugler and colleagues) already attributed only a modest share of the variation to de novo mutations: 2.6%.
Many genes still to be found
You may have read that roughly 25% of genes "strongly linked" to autism remain undiscovered. The phrase can be confusing: in the modelling preprint, that quarter refers only to genes with a very strong effect (likelihood multiplied by more than 20), of which 35 out of an estimated 46 would already be known. For genes with a more moderate effect (likelihood more than doubled), the identified genes would represent only about 21%, or 126 of about 601 according to the model.
Discovery would also be much faster if studies included autistic people with an intellectual disability or developmental delay. A limit the authors note: genes found this way may mostly reflect certain profiles.
Anxiety, depression, ADHD: different genetic roots
The second piece of work is published in a peer-reviewed journal: Molecular Psychiatry, online on 12 September 2026, publicised in late October (Adeniran Okewole, Varun Warrier and colleagues, University of Cambridge). It draws on the US SPARK cohort: clinical data from 74,204 autistic people, of whom 17,582 had genetic data. Across the whole group, 56% had at least one co-occurring psychiatric condition.
The researchers grouped these conditions into three dimensions: behavioural (ADHD, disruptive behaviour disorders), emotional (anxiety, depression) and thought disorders (schizophrenia, bipolar disorder). Results:
- each dimension is partly heritable, through common variants;
- the autism genetic score is not associated with any of the three dimensions;
- genetic scores for ADHD and depression, built in the general population, are associated with them;
- rare variants are linked to intellectual disability, not to these three dimensions.
In other words, among the autistic people studied, ADHD, anxiety or depression seem partly linked to the same genetic factors as in the general population, rather than to the genetic factors of autism. This is an association seen at group level, not a prediction for any one person. These conditions are not simply "part of autism": they deserve to be recognised and supported in their own right.
Limits: a single cohort of volunteer families, diagnoses not checked against medical records, genetics limited to people of European ancestry, and co-occurring diagnoses that may be missed in carriers of rare variants, who often have high support needs. The study does not explain why these conditions are more common in autistic people. Funding: including doctoral studentships (Cambridge Trust, Trinity-Henry Barlow studentship), the Wellcome Trust and the Simons Foundation (SFARI); no competing interests declared.
What this means for you
Autism is diagnosed through a clinical assessment, not through DNA. The UK NICE guidelines say not to use genetic tests routinely as part of an autism diagnostic assessment, in children as in adults, and to consider them case by case: specific physical features, congenital anomalies, intellectual disability. For the American College of Medical Genetics and Genomics (ACMG), genetic evaluation comes after a reliable diagnosis, to look for a cause, inform the family and guide care.
This evaluation can identify a syndrome that needs specific medical follow-up, such as Phelan-McDermid syndrome. If you are wondering about this for yourself or your child, talk to your doctor.
If you are wondering instead whether you have autistic traits, a self-assessment questionnaire such as the AQ-10 or its longer version, the AQ-50, can help you prepare a conversation with a professional. It is not a diagnosis.
Read also: hundreds of genes, two brain states in mice, and four autism profiles in 5,000 children.
Frequently asked questions
Is autism genetic?
Largely: heritability is estimated at about 80%. But for most autistic people, it is not about a single gene. It is the sum of many small common variations, sometimes with a rare variant, and with non-genetic factors.
Is there a genetic test to find out if you are autistic?
No. A genetic evaluation may be offered after diagnosis, in certain situations, to look for a cause. It does not make the diagnosis.
Does carrying a variant in one of these 253 genes mean you are autistic?
No. These variants raise the likelihood without making it certain: according to a model estimate, a typical mutation of this kind would lead to autism in only a little more than one carrier in two.
Do these discoveries mean a treatment is coming?
No. They describe genes and biological mechanisms. They may mostly help to better follow people who carry certain rare variants, who sometimes have other health needs.
References: Satterstrom FK, Auwerx C, Fu JM, et al., Autism Sequencing Consortium, Talkowski ME, Daly MJ (2026). Rare variation illuminates the distinct and pleiotropic genetic architecture of autism across neuropsychiatric traits. medRxiv, preprint not peer reviewed, posted 26 August 2026. doi:10.64898/2026.08.24.26360398. Rare coding variants, 62,429 autistic and 33,316 non-autistic individuals. | Nadig A, Fu J, Satterstrom FK, Auwerx C, Zhang Z, et al., Talkowski M, Robinson EB, O'Connor LJ (2026). Estimating the contribution of coding mutations to autism. medRxiv, preprint not peer reviewed, posted 27 August 2026. doi:10.64898/2026.08.25.26361328. Modelling study, 38,680 trios. | Smith RL, Liao Y, Klei L, et al., Gandal MJ (2026). Autism genes converge on three functional programs organized by neuronal subclass, developmental timing, and cortical patterning. bioRxiv, preprint not peer reviewed, posted 23 September 2026. doi:10.64898/2026.09.22.753508. | Okewole A, Bourque VR, Koko M, Huguet G, Borglum AD, Grove J, Jacquemont S, Baron-Cohen S, Warrier V (2026). Common genetic variants are associated with increased likelihood of latent co-occurring neurodevelopmental and mental health factors among autistic individuals. Molecular Psychiatry, published online 12 September 2026. doi:10.1038/s41380-026-03880-1. Observational genetic study, SPARK cohort, 74,204 autistic people (clinical data), of whom 17,582 with genetic data. | Fu JM, Satterstrom FK, Peng M, et al. (2022). Rare coding variation provides insight into the genetic architecture and phenotypic context of autism. Nature Genetics 54(9):1320-1331. doi:10.1038/s41588-022-01104-0. Cohort of 63,237 individuals. | Zhou X, Feliciano P, Shu C, et al. (2022). Integrating de novo and inherited variants in 42,607 autism cases identifies mutations in new moderate-risk genes. Nature Genetics 54(9):1305-1319. doi:10.1038/s41588-022-01148-2. | Bai D, Yip BHK, Windham GC, et al. (2019). Association of Genetic and Environmental Factors With Autism in a 5-Country Cohort. JAMA Psychiatry 76(10):1035-1043. doi:10.1001/jamapsychiatry.2019.1411. Population cohort, 2,001,631 children including 22,156 autistic. | Gaugler T, Klei L, Sanders SJ, et al. (2014). Most genetic risk for autism resides with common variation. Nature Genetics 46(8):881-885. doi:10.1038/ng.3039. | Schaefer GB, Mendelsohn NJ, Professional Practice and Guidelines Committee (2013). Clinical genetics evaluation in identifying the etiology of autism spectrum disorders: 2013 guideline revisions. Genetics in Medicine 15(5):399-407. doi:10.1038/gim.2013.32. | NICE. Autism spectrum disorder in adults: diagnosis and management (CG142), recommendations 1.2.11 and 1.2.17, last updated 2021. | NICE. Autism spectrum disorder in under 19s: recognition, referral and diagnosis (CG128), recommendation 1.7.1, last updated 2017. | Barker H. Number of autism-linked genes more than triples in new analyses. The Transmitter, 8 October 2026.