5 ms·

Can anyone comment on privacy for this company? I know that's an issue around 23&me

craftyguy 8y ago
It's a serious issue for any service that offers to sequence any/all of your genome. AFAIK, none of them get it right. I hope to be corrected here, but I doubt it.
jghn 8y ago
I believe you can opt out of them storing your data once you retrieve it.
- dsfyu404ed 8y ago
  Which is useless unless we can get someone on the inside to vouch for the fact that it's actually not stored
  - stochastic_monk 8y ago
    Most companies at least track frequencies. We used to think that deidentified was enough, but apparently it’s not that hard to figure out who a person is if you know that much about them. (Especially if you have relatives.) There’s been a lot of work with homeomorphic encryption, but I think that the new avenue of research using sketches to provide a sense of anonymity while preserving comparability.
  - jghn 8y ago
    The storage of the data is a not inconsequential portion of the total cost. A lot of work is done to cut costs on genomic data storage. No better way than not storing it in the first place
  - tempestn 8y ago
    They explicitly state they're GDPR compliant, so at the very least if they keep customer data after saying they've deleted it and are discovered they would probably be looking at huge fines. Doesn't make much sense from a business perspective, especially given that most customers probably won't bother to have their data deleted.
striking 8y ago
Link to privacy policy: https://us.dantelabs.com/pages/privacy-policy https://us.dantelabs.com/pages/privacy-policy
FriedPickles 8y ago
I believe in the future it will be basically free to sequence the genome from a strand of hair. Given that, is it reasonable to try protecting your genome data?
- mattnewton 8y ago
  Imo yes; the ability to match dna to individuals without their consent should belong to law enforcement only. I am holding out hope for legislation that makes this work.
  - conception 8y ago
    The ability to match dna to individuals is sadly pretty easy and even the idea that you can "de-identify" dna is the elephant in the room for things like HIPPA since dna is the most identifiable thing a person has. Looking at things like this - https://phys.org/news/2013-03-easy-identity-cell.html https://phys.org/news/2013-03-easy-identity-cell.html https://33bits.wordpress.com/about/ https://33bits.wordpress.com/about/ etc Having DNA and just looking at the traits that person has will reduce the population down to a pretty manageable number. Any other bits - location, gender, etc will be enough to make an educated guess to an almost certain one. This will not become more difficult to do in the future.
    - echelon 8y ago
      I wouldn't be surprised if we trained ML models to map DNA to adult faces. It doesn't seem out of the realm of possibility.
      - m1573rp34130dy 8y ago
        there is a thing called developmental noise, thats the outcome of phenotype after post expressional events further modify the expression products [proteins, RNA, MeDNA etc.] the face you interpolate from genetic sequence will be an approximation.
      - JProthero 8y ago
klmr 8y ago
> I know [privacy is] an issue around 23&me In what regard? I’ve not heard of any privacy related issue with 23andme, and I follow the field quite closely (professional interest). On the contrary, 23andme has been known to uphold customer privacy in the face of government requests [1]. The FTC is investigating 23andme and other companies (which is a good thing), but there is no indication that they’ve found any violation. [1] https://www.23andme.com/transparency-report/ https://www.23andme.com/transparency-report/

It's a serious issue for any service that offers to sequence any/all of your genome. AFAIK, none of them get it right. I hope to be corrected here, but I doubt it.

jghn 8y ago

I believe you can opt out of them storing your data once you retrieve it.

dsfyu404ed 8y ago
Which is useless unless we can get someone on the inside to vouch for the fact that it's actually not stored
- stochastic_monk 8y ago
  Most companies at least track frequencies. We used to think that deidentified was enough, but apparently it’s not that hard to figure out who a person is if you know that much about them. (Especially if you have relatives.) There’s been a lot of work with homeomorphic encryption, but I think that the new avenue of research using sketches to provide a sense of anonymity while preserving comparability.
- jghn 8y ago
  The storage of the data is a not inconsequential portion of the total cost. A lot of work is done to cut costs on genomic data storage. No better way than not storing it in the first place
- tempestn 8y ago
  They explicitly state they're GDPR compliant, so at the very least if they keep customer data after saying they've deleted it and are discovered they would probably be looking at huge fines. Doesn't make much sense from a business perspective, especially given that most customers probably won't bother to have their data deleted.

striking 8y ago

Link to privacy policy: https://us.dantelabs.com/pages/privacy-policy https://us.dantelabs.com/pages/privacy-policy

FriedPickles 8y ago

I believe in the future it will be basically free to sequence the genome from a strand of hair. Given that, is it reasonable to try protecting your genome data?

mattnewton 8y ago
Imo yes; the ability to match dna to individuals without their consent should belong to law enforcement only. I am holding out hope for legislation that makes this work.
- conception 8y ago
  The ability to match dna to individuals is sadly pretty easy and even the idea that you can "de-identify" dna is the elephant in the room for things like HIPPA since dna is the most identifiable thing a person has. Looking at things like this - https://phys.org/news/2013-03-easy-identity-cell.html https://phys.org/news/2013-03-easy-identity-cell.html https://33bits.wordpress.com/about/ https://33bits.wordpress.com/about/ etc Having DNA and just looking at the traits that person has will reduce the population down to a pretty manageable number. Any other bits - location, gender, etc will be enough to make an educated guess to an almost certain one. This will not become more difficult to do in the future.
  - echelon 8y ago
    I wouldn't be surprised if we trained ML models to map DNA to adult faces. It doesn't seem out of the realm of possibility.
    - m1573rp34130dy 8y ago
      there is a thing called developmental noise, thats the outcome of phenotype after post expressional events further modify the expression products [proteins, RNA, MeDNA etc.] the face you interpolate from genetic sequence will be an approximation.
    - JProthero 8y ago

klmr 8y ago

> I know [privacy is] an issue around 23&me In what regard? I’ve not heard of any privacy related issue with 23andme, and I follow the field quite closely (professional interest). On the contrary, 23andme has been known to uphold customer privacy in the face of government requests [1]. The FTC is investigating 23andme and other companies (which is a good thing), but there is no indication that they’ve found any violation. [1] https://www.23andme.com/transparency-report/ https://www.23andme.com/transparency-report/