Comment by occamschainsaw | Hacker News Reader

Comment by occamschainsaw | Hacker News Reader

I am a South Asian. It told me I was of South Asian ancestry. Was hoping for something more specific than that.

I love me some coffee (drink more than 3 cups a day), and it said I am very likely to consume less caffeine. It also told me I have a high chance of having smooth hair, but I have extremely curly hair.

And I am extremely skeptical of their traits in general. Some of the traits include: "cilantro taste aversion", "misophobia" (likely to hate chewing sounds), "ability to match musical pitch", "fear of heights" etc.

I am guessing they find correlations between survey answers and presence of genetic markers. A lot of these traits would also be affected by nurture not nature.

Overall, I am not happy with the experience. I don't know if it is the current state of technology or regulatory concerns holding them back from releasing better results. Another reason for their lack of specificity could be the absence of South Asian data in their datasets. Most of the disease diagnosis is targeted towards people with Ashkenazi Jewish descent. Probably because they are much more present in their datasets.

Either way it seems like a lose-lose for me because on one hand I am getting mediocre results, but on the other I am relinquishing my privacy.

Your comment made me go back to articles on data sharing 23andme does with companies. At first I read your statement about cilantro but it seems these are just the type of questions companies are paying big money for.

>For now, though, 23andMe’s biobank is the world’s largest repository of DNA samples that also contains extensive health information, willingly provided by customers who answer survey questions like “Do you like cilantro?” and “Have you ever had cancer?” 23andMe says its customers supply it with as many as two million new facts each week.

There’s only one problem—depending on your own ethics—and that is your data is helping pharmaceutical companies develop new drugs.

The continuation from the above quote:

>These surveys are proving valuable to drug investigators. This year the company found genetic variations strongly linked to whether customers consider themselves early risers, offering a clue about how to develop drugs that modulate alertness. [1]

[1] https://www.technologyreview.com/s/601506/23andme-sells-data...

That article was pretty helpful. Now I am entering the rabbit hole of reading about 23andme data sharing.

What strikes me as interesting is their current business model: customers are paying them and answering their survey questions. That is quite brilliant, they get specifics about the people in their existing dataset, and once a critical statistical mass has been reached, they can find correlations between traits and genomic data.

I would be happy to voluntarily provide my time, effort and data towards medical research. While I don't have ethical problems right now (I don't expect a lot of privacy anyways), there is a small nagging fear that these things could bite me 20 years down the line when these genomic tests are much more sophisticated and I have already given them my genome. Small benign hypothetical example: I start seeing ads about cilantro on Facebook.

That fear was overpowered by my curiosity to learn about my genome. That's on me.

> Another reason for their lack of specificity could be the absence of South Asian data in their datasets.

In 2016, there was a 100:1 ratio for Europeans vs East Asians in their datasets. I don't know how it was for South Asians, and I imagine it's probably somewhat better now, but I think this is likely a big factor.

My friend is a PhD in quantitative genetics and says that 23andMe is basically a shitty horoscope that costs some of your money and all of your privacy.

I can't speak to most of the things you list, but there is at least some evidence for genetic influence on cilantro preference (via olfactory receptors).

https://www.nature.com/news/soapy-taste-of-coriander-linked-...

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3589946/

Reply on news.ycombinator.com