Why care about genetic privacy?
Gene linkage in a small part of a genome
I’m keen to get my genome sequenced. This is so I can dig around in my data, snooping for clues about genealogy, psychology, and health. It’s a highly novel kind of narcissism.
I’m mostly just waiting for the cost to plummet some more. One reason it isn’t cheap yet is regulation. Most of this concerns data privacy, including my right to be protected from misinterpreting information I ask for and pay for.
I don’t necessarily understand my society’s extreme caution in this manner: I harbour an ambition to put my genome up online as a joke: so I can say “I am open source”. But some people have already done this for serious reasons, to find unknown relatives.
Here are some things you might worry about if you did such a thing.
As a European, I found it hard to believe how callous American healthcare is. They absolutely do use all kinds of data sources to limit coverage, ramp up cost, or exclude people entirely. Our current knowledge of genetics is not so powerful at predicting your phenotypical health, but many people have obvious risk markers, and it's only going to get better/worse.
Risk applies to: People with known pathological alleles living in non-single-payer countries.
Risk to me: Low.
If you are ever in a position where someone needs to make you look bad - e.g. a very fierce job competition; an election campaign; a custody battle - then any genetic frailty could be used against you.
Risk applies to: Extraordinarily ambitious people with known pathological alleles.
Risk to me: Low?
A flawed aid to police.
Having your genome fully available would open you to a small risk of being falsely implicated by a DNA test error or fragment collision. Note that, to avoid this completely, you'd need to not get sequenced by anyone else at all: police can subpoena genetics companies. (And total aversion is almost certainly not the highest expected value option.) This is in the "actual threats" section because the Golden State Killer was caught by uploading a crime-scene sample to an open genome database his relatives had used.
Risk applies to: Criminals and civil disobedients. Anyone living in an authoritarian regime with unpredictable antipathies.
Given a genome, it is not impossible for someone to grow tissue and like plant it at a crime scene. But it's going to be a damn sight simpler to just get some hair from my desk or whatever.
Risk applies to: Anyone who pisses off rich criminals, spies, police.
Risk to me: Low.
- Releases info about my family members.
The same principle that makes paternity tests so good means that when I disclose my genome, I give very strong indications to the genome of my family members.
Risk applies to: Any of my antecedents or descendents who fall into any of the above categories.
People knowing things about me is creepy!
Please don't look into Acxiom or Experian; it will only upset you.
Risk applies to: People obsessed with anonymity.
Homosexuality has a strong genetic component; this regularity means we will eventually develop a polygenic score to predict it, if noisily. 1 In a large chunk of the world, this is still not information that you can afford to make public.
(I was going to put this risk in the 'actual threats' bit, but the recently announced predictive algorithm is based on a ludicrously tiny sample size, and was built with forking paths and was probably overfit to hell.)
Risk applies to: Closeted gay people, and straight people who happen to score highly on what will always be a probabilistic score.
Risk to me: Could be.
Once a majority of people are sequenced, we could easily see customs about the genetic quality of partners arise. We all already do this unconsciously, using proxy variables like facial symmetry and sheer conspicuousness. I would expect genetic pre-nups, at least among the kind of people who have pre-nups today.
(You might think that postwar Europeans are completely inoculated against such things; see Iceland for how quickly this has already taken off.)
(You might think that no one you loved would ditch you over some bad genes. After all, there's adoption! But most people are scary serious about children: they'll do terrible things in their name.)
Risk applies to: Anyone who wants kids, or wants to be with someone who does.
Controversy: People really seem to hate clones for no good reason. (It threatens their sense of self, or growth mindset, or divinity.)
Harm to me: None; more of me is a good thing. I am a fragile combination of organic molecules; if someone wants to give me a backup, I am happy with this.
I guess they could torture the clone to get to me? But this is silly and I do not negotiate with silly terrorists: it would be millions of times cheaper to just abduct my lover or something. But people in general sure are crazy about those genetically similar to them.
Probability: Of me being cloned off some ancient databank? Miniscule.
- Tailored bioweapons.
You might worry about being assassinated with a virus which only kills you. Dunno why this would worry you more than very real weaponised things which would kill everyone around you too, seriously what is the matter with you?
Harm to me: No. I am unlikely to sufficiently piss off anyone with a spare billion dollars.
(Then, of course, there’s everything I haven’t thought of - and everything no-one has thought of - everything which the great random-situation generator called physics hasn’t yet thrown up. You should probably seek privacy based on this, more than the above.)
The size of the genetic component seems to be 30% - 40% of variance. A further epigenetic component is being investigated right right now.
Current polygenic scores for complex traits like educational attainment or symptomatic psychosis are about r = 0.3, but note that this is an incredibly young methodology: the oldest citation on that Wiki page is 2009.
"String" commented on 26 August 2017 :