Summary: Whole-genome sequencing from a saliva sample now costs a few hundred dollars and yields a file of a few gigabytes that you keep and can query forever, instead of re-running targeted genetic panels as research expands. With LLMs writing the analysis scripts, anyone can ask questions of their own DNA. Beyond curiosity (eye color), it can surface clinically useful facts, especially drug responses (pharmacogenomics), that would otherwise be found by trial and error.
Key ideas
- Prenatal genetic panels test specific mutations, and the list grew from ~6 to 30–40 tests between two children. Since DNA doesn’t change, sequence once and re-query as research advances, “like SQL” (explainable—152-biohacking-like-a-data-scientist).
- Cost: ~$300 per person; the host sequenced his whole family (explainable—152-biohacking-like-a-data-scientist).
- Findings that mattered: statins (often given preventively in Israel from age 50) are toxic for him; alcohol sensitivity tied to liver-related variants; hyper-metabolism of a common dental anesthetic, explaining why he always needed large doses (explainable—152-biohacking-like-a-data-scientist).
- Many responses to ADHD medications are genetically encoded, yet medicine still titrates by trial and error (explainable—152-biohacking-like-a-data-scientist).
- Single-gene explanations can mislead: a relative’s child’s deafness turned out to have a different cause than the family mutation (explainable—152-biohacking-like-a-data-scientist).
- Use LLMs (Claude, ChatGPT, Gemini) to write Python scripts over the genome file and to interpret reports before seeing a doctor (explainable—152-biohacking-like-a-data-scientist).
Disagreements & open questions
Takeaways
- Consider whole-genome sequencing (~$300) and check pharmacogenomic variants (e.g. statins, anesthetics) before you need those drugs (explainable—152-biohacking-like-a-data-scientist)