Standard genetic testing for autism spectrum disorders identifies the underlying cause in only about 25% of cases, leaving thousands of families without answers. This gap may partly stem from the limitations of conventional short-read DNA sequencing, which struggles to detect complex structural variants and fails to capture the full diversity of human genetic architecture. Advanced long-read sequencing combined with pangenome reference maps offers a promising solution to this diagnostic blind spot.
Researchers analyzed 189 individuals from 51 families with previously unexplained autism cases using long-read sequencing to construct near-complete genome assemblies. This approach generated exceptionally high-quality data with average contig lengths of 43 million base pairs. The team applied both read-based and assembly-based strategies to comprehensively characterize de novo mutations, structural variants, and DNA methylation patterns. Using pangenome controls, they efficiently filtered out over 97% of common structural variants, allowing focus on potentially pathogenic changes.
The analysis revealed three confirmed pathogenic variants in established autism genes TBL1XR1, MECP2, and SYNGAP1, plus nine additional candidate variants that were predominantly missed by standard short-read sequencing. Interestingly, while no increased burden of structural variants was observed on autosomes when comparing affected individuals to unaffected siblings, there was a suggestive trend toward increased structural variant burden on the X chromosome among affected females. This finding aligns with emerging evidence that X-linked variants may contribute more significantly to autism in females than previously recognized, potentially explaining part of the observed sex bias in autism diagnoses and highlighting the importance of comprehensive genomic analysis in clinical settings.