AI and Fruit Fly Tests Link BRSK1 Gene to Rare Neurodevelopmental Disorder

Researchers at Baylor College of Medicine and the Duncan Neurological Research Institute at Texas Children's Hospital have identified mutations in the BRSK1 gene as a primary cause of a rare, complex neurodevelopmental disorder. Published in the American Journal of Human Genetics, the study leverages artificial intelligence and cross-species modeling to bridge diagnostic gaps for medically underserved populations. The initiative originated through the Texome Project, which provides free genomic testing to families in Texas facing unexplained rare conditions. Initial standard genetic screenings of a participating parent-child pair yielded no results until the AI-MARRVEL tool flagged a rare variant in BRSK1. Subsequent collaboration via GeneMatcher connected the research team to six additional families, expanding the cohort to ten affected individuals. Clinical observations across the cohort reveal a spectrum of developmental delays, including speech and language impairments, intellectual disability, autism spectrum disorder, attention-deficit/hyperactivity disorder, anxiety, hypotonia, microcephaly, and seizures in select cases. Notably, symptom severity varied significantly even among family members sharing identical variants, indicating pronounced variable expressivity. To validate these clinical findings, the researchers utilized Drosophila melanogaster as a model organism. The fruit fly equivalent of BRSK1, known as sff, demonstrated nearly identical neural expression patterns in humans. Genomic suppression of the sff gene resulted in pronounced motor deficits, stress-induced seizure-like activity, heat susceptibility, and reduced lifespan. Introducing the standard human BRSK1 gene successfully mitigated these neurological impairments, confirming functional conservation across species. Conversely, deploying patient-derived variants only yielded partial functional recovery, establishing that the mutations compromise but do not entirely abolish protein activity. Subsequent cellular analysis indicated that diminished BRSK1 function disrupts microtubule organization, a critical structural framework for neuronal development and synaptic communication. The study underscores the transformative potential of integrating machine learning algorithms with experimental biology to decode undiagnosed genetic conditions. By coupling AI-driven genomic interpretation with targeted model organism validation, the research not only provides definitive answers for previously undiagnosed families but also establishes a reproducible framework for identifying novel neurodevelopmental disease mechanisms. Lead investigators emphasize that this methodology will continue to expand access to precision genomics for historically marginalized patient populations, accelerating both clinical diagnosis and therapeutic target discovery.