Genes are probability, not destiny

By Razib Khan | April 4, 2012 10:53 pm

On my Facebook feed some geneticist friends of mine were passing around an article in The New York Times, Study Says DNA’s Power to Predict Illness Is Limited. The article is based on a paper, The Predictive Capacity of Personal Genome Sequencing. In its turn this paper, as well as the write up in The New York Times, has been widely criticized in the genomics and genetics community. On the other hand, Luke Jostins cautions that perhaps one lesson statistical geneticists should take away from this is that they should do a better job communicating to the public.

I’m with Luke in spirit. But I wonder: how much more difficult is it to think in terms of probabilities rather than binary outcomes? Because that is the ultimate issue. Even geneticists sometimes talk in “nature vs. nurture,” as if the two are oil and water poured in exact ratios into a basin. Of course the reality is that nature and nurture interact, and one shouldn’t discount either. Instead of talking about disease, perhaps it might help to focus on a quantitative trait like height. This is ~90 percent heritable. That is, ~90 percent of the phenotypic variation can be explained by genetic variation.

The fact that Kobe Bryant’s father was a journeyman professional basketball player no doubt resulted in his son’s sights being set rather high in that arena. That is nurture. But as it happened, Kobe also inherited height from his father. That is nature. Talking about the two as if they are at cross-purposes misleads.

Recall that height is ~90 percent heritable on the population level. But it turns out that the standard deviation of identical twin height differences is still ~35-40 percent that of random siblings! What I want to see next, an article in The New York Times: “Identical twins not always identical in height; genes don’t explain everything.”

CATEGORIZED UNDER: Human Genetics
  • phanmo

    “…his son’s sites being…”
    It took me a couple of seconds… you means “sights”, right ?

  • Eric

    There are definitely more “talented” players than Kobe, though not many (Lebron and Kevin Durant come to mind), but I’d still say he’s better than them because he is more “skilled”. He has a better mastery of the game.

    Of course you need both skill and talent to make it in the NBA, but which is more important?

  • Juan

    While we are on the subject of genes as destiny, I want to know what people here would make of this blog post:

    http://neuroskeptic.blogspot.co.uk/2012/03/personality-without-genes.html

    According to this blogger, a meta-analysis done a while ago where people of European descent were tested for possible correlations between a set of psychological traits and SNPs. Surprisingly, no substantial associations were found!

  • Kieron

    “even the horse with the highest odds can lose the race”

  • marcel

    Phanmo @1: If only you moved your lips when you read, just enough to sound out the words in your head (like I do), it would have taken you much less than a couple of seconds!

  • Charles Nydorf

    The argument about the human capacity to manage probabilistic thinking is itself entangled with nature/nurture debate. Kahnemann and Tversky claim that humans are hardwired to be bad a probilistic reasoning while Gigerenzer maintains that the problems are the result of cultural conditioning and can be overcome by education. I tend to think that Gigerenzer is closer to the truth.

  • ytheleus

    “Identical twins not always identical in height; genes don’t explain everything.”
    Yes, but it doesn’t follow that the remaining 10% percent variation (with respect to height) is therefore non-biological. The remaining 10% presumably in the case of height has to be biologically determined. The same logic might not necessarily apply to other traits of course like intelligence but it could also.

  • Chris

    I heard this analogy from a friend and was wondering your thoughts.
    DNA is like the CPU and epigenetics is like the software.

  • http://www.michaeleisen.org Michael Eisen

    You should write a paper containing a simple mathematical model of the relationship between genometype and height, apply it to previously published twin height data, and press release the shit out of it saying that “Parents should not waste their money sequencing their kids genomes since it won’t tell them whether their kids will be basketball stars” and you’re set.

  • Sandgroper

    #6 – My experience (lots of it) talking to people about geophysical hazards is that the default option is for people to want to think about hazards in binary terms, but if you tell them that they need to think of them in probabilistic terms, and if you give them ‘familiar’ analogies, they get it instantly.

    Times I have talked to family/friends who have been conditioned by decades of nature/nurture debate, when I say “complex interaction of genes and environment”, likewise.

    So I think you are right.

  • http://blogs.discovermagazine.com/gnxp Razib Khan

    Of course you need both skill and talent to make it in the NBA, but which is more important?

    is the question even on point? all NBA players need skill and talent, necessarily (even manute bol had to develop particular skills).

    DNA is like the CPU and epigenetics is like the software.

    don’t get how DNA is like CPU.

  • pconroy

    @8 Chris,

    Here we go again, epigenetics is the new hope of the blank slaters…

    IMO, a more apt metaphor would be that Epigenetics is like the Lakers jersey Kobe wears

  • Eric

    @Razib

    Definitely! You have players like Tracy McLazy, Carmelo Anthony etc. with insane amounts of talent, but slack off in refining certain aspects of their game. Defense in both cases, for Tracy McGrady you can also add inability to fit into a system. Take T-mac out and the team is better, unlike Melo on the Nuggets.

    Needless to say there are also people who will never be S-Class players no matter how hard they try. Lol, Manute Bol, haven’t heard that name in a while.

    @12

    Definitely not a ‘blank slater’, but I think your comparison is over doing it as well.
    Have you seen this yet?
    http://www.nature.com/scitable/topicpage/obesity-epigenetics-and-gene-regulation-927

  • ytheleus

    Actually height is subject to a lot of environmental influence under certain conditions. Look at photos of Dickensian London and see the height disadvantaged aplenty. I believe this is due to being very ill during growth spurts with the result that the spurt is halted and that growth opportunity is lost forever.
    Barring such environments (with poor sanitary conditions etc) then it would be 90% heritable.

  • Alam

    “DNA is like the CPU and epigenetics is like the software.”

    I think DNA is like firmware for the devices. You can have different applications behaving differently using the same device and firmware :)

    Its like using Skype or Facetime on iPad. Facetime might have some advantage over Skype, because its internal product. Both products were developed in diff environments, but on the same hardware and firmware !

  • Chris T

    3 – There are probably so many genes influencing personality that the effect of any one SNP is tiny.

    14 – Heritability estimates are actually pretty meaningless without environmental context.

    Post topic – Humans are really bad at handling abstractions like probability. It’s very difficult to intuitively grasp large numbers, but we instinctively empathize with a person who’s suffered the effects of a rare event. Thus, when it comes to a disease, we can readily imagine suffering from it and hearing that we’re genetically more likely (even if marginally) sounds like a death sentence.

  • Sandgroper

    Human attitudes to risk are well researched and documented. An involuntary risk is 2 to 3 orders of magnitude less tolerable than a voluntary risk. Add additional factors for ‘unknown’ and ‘dread’ events which are less tolerable, and it is not difficult at all to see why someone might obsess about a probability of suffering a debilitating disease that is orders of magnitude less than the probability of being killed or seriously injured in an automobile accident, for example, which is perceived to be a familiar and voluntary risk over which the person has (rightly or wrongly) some level of personal control. A common yardstick used by risk analysts is the voluntary risk tolerated by smokers – they understand the risk, and that it is actually very high compared to the risk they bear from numerous other hazards, but they tolerate it.

    That doesn’t mean they are ‘bad at handling the abstraction’, it means they have different attitudes to the risk levels from different types of events. It is not hard at all for someone to grasp a large number if you relate it to the probability of occurrence of an event they have some familiarity with.

  • Denise

    Genes are probability for SOME people. Stephen Hawkings did not become serious about his work until he was stricken with his disease. He may not have ever done what he’s done if he had remained healthy.

    Genes ARE destiny, for most. Could Trayvon Martin EVER have accomplished Hawking’s deeds?

  • Justin Loe

    #17: Without sidetracking these comments, my reading (and one of my major reasons for exiting the social sciences) was that most of the foundational assumptions of rational choice theory were false.

    I think there’s good evidence that most human beings are quite poor at handling abstraction, in my view. (with respect to irrationality, we’re all significantly less rational than would be predicted by the econometric models widely in use, though not by the recent behavioral economics findings).

  • Chris T

    17 – Everything you listed is exactly why humans are bad at handling abstractions. We discount threats we have some input (driving) even when the risks are much higher than other risks.

  • Sandgroper

    Well, the argument is that if people were rational, and if they understood the numbers, they would be most concerned about the highest risks in absolute terms.

    But there are a number of things to consider:
    1. When it is explained to people, they trend towards being more rational about it, i.e. they are capable of getting it.
    2. People will argue that it is perfectly rational for them to be much more accepting of voluntary risks – they are looking at it from the perspective of perceived benefit.
    3. It might be very difficult to further reduce risk from say automobile accidents, and that might be regarded as too high a price to pay, in terms of loss of convenience, efficiency, etc, whereas minimal investment could substantially reduce an already lower level of risk from another hazard, in which case it should be done. People expressing concern about the lower level risk is really reflecting that – they are looking at risks in aggregate and saying do what you can in practical terms to reduce what you can, without paying too high a price in terms of cost/benefit or imposing too much inconvenience or too many restrictions on us.
    4. People might not regard all forms of death, say, as equal – and it’s hard to argue they are not being rational about that, it’s their choice.
    5. I think where people are somewhat irrational is in their aversion to multiple fatality events (e.g. a single event that kills 2 people is perceived to be worse than two separate events that kill one person each, although the outcome is the same), but that doesn’t apply in the case of genes. In any case, I think that is conditioning.

    All I’m arguing for is that in the case of genes, I think Luke Jostins is right, and that it’s an effort that should be made – people can understand probability, in my experience, and they can understand a rational ranking of magnitudes of risk. They might not want to listen, because it makes them feel uncomfortable. They don’t like being told things are not binary; they are uncomfortable with uncertainty, but they get over it.

    Of course, I MIGHT not include #18 in that.

  • pconroy

    @21,
    In terms of driving, I expect that the risks will plummet as Google and others start rolling out self-driving cars, which can basically be programmed to avoid crashing.

    http://theweek.com/article/index/226533/the-week-contest—future-cars

    Of course, there will be a refining period where the risk might go up for a few years, initially.

  • Sandgroper

    Yes, hopefully. The problem with traffic related risk is that people tend to see it as voluntary and something they have a measure of control over, but they may have less control than they think – i.e. you can’t control what the other driver does. Plus people generally over-rate their own driving skills. So people tend to underrate the risk and may be too tolerant of it. This is not hard to observe in reality.

  • Chris T

    Two counter points:
    1. People tend to overestimate the amount of control they actually have.
    2. People also overestimate their own competence.

  • Unscrewed

    There’s one more reassuring thing about risks that are voluntary: to some extent you can adjust the odds to suit your own risk tolerance. No, you can’t control the other driver, but y0u can stay off the roads on New Year’s Eve when he’s likely to be at his drunkest.

    If this kind of piddling safety measure reassures you, is your sense of probability faulty? Not necessarily. Humans are strongly motivated to exert control over their own fate–it’s a well-known finding in psychology that having an external locus of control sets you up for all kinds of neurosis. The same level of perceived risk feels so much worse when there’s no way to satisfy the drive for control it creates.

  • Sandgroper

    Agreed.

    The other frustrating thing is if you can find a way to reduce the risk people are exposed to, they will respond by increasing risk-taking behaviour in order to increase their risk again.

    But I don’t really see that applying to genes. Well, I could foresee situations where it might.

  • Mikey Mike

    Well, I do agree that for most features, your genes are just probability (like probability of getting rheumatoid arthritis or something like that), but it is destiny for certain features, like if you’re homozygous for cystic fibrosis or something like that. There’s an incredible amount of randomness in heredity and also in brain development.

    Your choice of showing a picture of Kobe is interesting because there have been a few twins in the NBA and other sports. I think that enough twins have made it that it’s high probability of destiny.

    In the NBA, there have been Horace and Harvey Grant, Brook and Robin Lopez, Jarrod and Jason Collins, Markieff and Marcus Morris
    MLB: Jose and Ozzie Canseco, one on PEDs, the other without or with less.
    NFL: Ronde and Tiki Barber

    I’m sure that there are other examples, but I only follow the NBA closely so I can’t think of any other ones.

  • Eric

    I think Robin got by purely on the “probability” that he might turn into Brook.

  • Sandgroper

    #25 – Yeah.

    One of the big benefits I observed from educating lay people about risk was, not only did they grasp probability reaonably well when it was explained, but having an understanding of the way that the hazards are manifested, and things they could do to minimise their exposure to the hazards, made them feel a whole lot better. They flipped and started asking about OK what can WE do to reduce the risks we are exposed to, and the whole discussion went from negative to constructive. That really helps a lot in trying to manage risks to a community, if they are prepared to learn how to change behaviour to minimise their own exposure.

    I had a colleague in another country who reported a lot less success, and some very much more intransigent people, but obviously not everyone is the same.

  • Sandgroper

    Unscrewed, that’s actually a very helpful observation – do you mind if I use what you’ve said? I’d like to pass it on to some people who can use it.

    I don’t want to be a pain in the arse, but if you had a good reference or two on that, it would be great. If not, never mind, it’s kind of a self-evident truth once it’s elucidated.

  • Violet

    I think people are good at risk evaluations and poor at probability evaluations. I do think it is hard to have an intuitive grasp of very low probabilities (when expressed as 0.1% or 0.001% rather than as odds) but once the consequences of the events are evaluated, most people have pretty good understanding if something is worth attention or not according to their own risk-tolerance strategies.

    Also, most common problem I find with probability presentation is the differentiation between total probability and conditional probability. Some conditional probabilities are expressed as if they are total probabilities and it is difficult to tease those out without domain-specific knowledge in some areas.

    So, yeah agreeing with comments above, people are good at understanding the odds if explained properly.

    #26, That is the most frustrating thing for all safety-training people I guess. Perhaps there is a distribution of risk-tolerance among the population that remains constant. If one part of it is changed, people automatically compensate by responding it opposite ways. At least, it shows people are constantly updating their priors.

NEW ON DISCOVER
OPEN
CITIZEN SCIENCE
ADVERTISEMENT

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at http://www.razib.com

ADVERTISEMENT

See More

ADVERTISEMENT

RSS Razib’s Pinboard

Edifying books

Collapse bottom bar
+

Login to your Account

X
E-mail address:
Password:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »