I asked AI what that means. Came up with a variant pathogenicity analysis and a new hypothesis.
Michael is 4. He doesn't hear well. Two broken copies of the STRC gene. One confirmed pathogenic. The other: "Variant of Uncertain Significance." Three words that block him from gene therapy trials.
I'm not a geneticist. I build websites, shoot video, and do AI education. I have an AI agent (OpenClaw, powered by Claude Opus 4.6) running on my laptop. It searches databases, downloads protein structures, runs analysis. I ask questions from my phone while Michael plays next to me.
One question led to reclassification evidence. Then conservation analysis. Then a hypothesis about fitting the gene into a single therapy vector. Then six structural experiments. Then three emails to the scientists who pioneered this research. One responded overnight. By day three, we had a new hypothesis: a self-dosing gene therapy where sound itself activates the treatment, backed by an ODE model showing therapeutic protein levels in 13 hours.
Science shouldn't be locked behind jargon. There's a podcast and a video below (both AI-generated) for anyone who'd rather listen than scroll through protein structures.
Egor and Michael, Hong Kong
Computational evidence supporting VUS to Likely Pathogenic reclassification for NM_153700.2:c.4976A>C p.(Glu1659Ala)
Computational hypotheses for accelerating STRC gene therapy. These require experimental validation.
Step-by-step methodology so anyone can reproduce these results