Single nucleotide polymorphisms predict symptom severity of autism spectrum disorder.

Jiao et al. (2012) built a 29-SNP genetic score to predict autism severity. They tested it against the Childhood Autism Rating Scale (CARS). The model got 67 % of cases right—better than chance, but far from perfect.

One SNP, rs878960 in the GABRB3 gene, kept showing up as useful on its own.

The 29-SNP panel gave a weak but real signal. It sorted kids into mild versus severe groups with 67 % accuracy. rs878960 alone was the strongest single marker.

In plain words: genes added a small piece to the severity puzzle, not the whole picture.

Sugie et al. (2005) already linked a different gene (5-HTTLPR) to drug response in autism. Yun extends that idea—genes can predict phenotype, but now for natural severity instead of medication change.

Kamp-Becker et al. (2010) showed Asperger and high-functioning autism are just points on one continuum. Yun gives that continuum a genetic ruler, backing up the “severity is continuous” view.

Sun et al. (2015) proved standard tools like ADOS work in China. Yun tries the same validation job for a DNA-based tool, but with weaker results—67 % versus Xiang’s over-80 % agreement.

Farrant et al. (1998) remind us that parents and siblings often carry tiny autistic traits. Yun’s severity model could, in theory, be run on undiagnosed relatives to see if the same SNPs track sub-threshold quirks.

For now, you still need ADOS, ADI-R, and good clinical judgment. Yet keeping an eye on GABRB3 and related SNPs may one day sharpen early estimates of how much support a toddler will need. Until larger studies boost accuracy above 90 %, treat genetic severity scores as conversation starters with families, not final verdicts.