Too Much Diversity

There’s a new paper out in Nature, by Wenqing Fu and many other people, about the recent origin of most variants in protein-coding genes. They conclude that most are less than 5-10,000 year old – younger in Europeans than in Africans.  This is a natural consequence of the shape of human demographic history – there was a huge population increase with the advent of agriculture, and more people meant more mutations.  That agricultural expansion happened somewhat earlier in the Middle East and Europe than in Africa.

More people means more mutations – that was clear to some of us five years ago.  We were primarily interested in the rate at which adaptive mutations were generated – that increase, coupled with the selective pressures associated with a very different agricultural way of life, had to materially speed up adaptive evolution in humans.  More so in some places than others, of course.

A very few mutations are beneficial, some are neutral and many are deleterious, although the degree of harm inflicted varies widely.  So the population expansion also increased the number of bad mutations – but unless  selection also relaxed, it would not have changed the per-capita number of deleterious mutations, or the distribution of their effects (what fraction had large, medium, or small effects on fitness).  It increased the diversity of deleterious mutations – they are more motley, not more common. The article never talks about that per-capita number, or, if it did ,  I was unable to winkle it out.  It talks about ages and numbers of mutations –  but not the mean number, in either of the two populations studied (European Americans and African Americans) .  I think it would been a lot clearer, confused fewer reporters, if it had made that distinction. On the other hand,  depending on the facts on the ground, talking about mutational load might be a grant-killer. There was a paper earlier this year  (with many of the same authors) that used about half of the same data and did mention per-capita numbers. I’ve discussed it.

Some of these deleterious mutations have negative effects large enough to be considered disease genes. You can imagine two extremes.  In one, everyone with schizophrenia has the same nasty mutation.  In the other, virtually every family with members suffering from schizophrenia has a different causal mutation.  That second pattern is bad news for medicine: it reduces the chance of finding a drug that helps a big fraction of sufferers. People talk about individualized therapy, but of course that’s impractical. Hard to test (do we create an army of clones for phase II?), impossible to share the  development costs among many customers.

Unfortunately, that second pattern is pretty close to reality.  Most of it has little to do with  anything that happened a few thousand years ago – any particular mutation with a strong enough effect to be much of a schizophrenia risk gene doesn’t last that long.  So a recent (Bronze-Age) increase in the diversity of deleterious mutations doesn’t make much difference.  We were already far from common-disease-common variant.  On the other hand, a population with more genetic load, more deleterious mutations (and not ones with incredibly small effects), might well be more susceptible to schiz. But that doesn’t point to a drug either: it merely says that it is better to have fewer holes in your genome.  That doesn’t suggest a therapy, at least not to me.

The paper says that there may be an excess of  weakly deleterious mutations in Europeans due to bottlenecks back in the Ice Age.  The idea works like this: selection is less efficient in small populations.  Deleterious mutations with an effect s < 1/Ne drift freely and are not efficiently removed by selection.   This effect takes on the order of Ne generations – so a population reduced to an effective size of of 10,000 for 10,000 generations ( ~250,000 years) would accumulate a large-than-usual number of deleterious mutations of effect size ~10-4.    Lohmueller et al wrote about this back in 2008:  the scenario they used had a European ancestral bottleneck 200,000 years long, which is A.  what you need to make this scenario work and B. impossible, since it’s way before anatomically modern humans left Africa.  Back to the drawing board.

About these ads
This entry was posted in Uncategorized. Bookmark the permalink.

9 Responses to Too Much Diversity

  1. nooffensebut says:

    To play devil’s advocate, didn’t Lohmueller et al find not only more mutations in Europeans but also more homozygosity of those? Granted it didn’t have this study’s power, but it doesn’t paint a picture of mere “motley” diversity.

    I thought a possible revision of the mutation rate based on empirical evidence might force us to set our clocks back such that each humanoid and race is twice as old as we thought.

    I was just reading about the COMT-marijuana interaction effect on psychosis. A pre-modern environment could shield gene-environment risk alleles from selection. It seems like there is a very strong desire on the part of many to hope that g-e interactions and non-SNP variation are trivial because they aren’t as straightforward to study as a large GWAS. Mutational load is not mostly beneficial, but we don’t really know how bad a given quantity of missense mutations or homozygous missense pairs is. It’s not as if determining that Europeans carried more genetic load would necessitate reevaluating the quantitative genetics of IQ.

  2. gcochran says:

    That Lohmueller paper looked at SNPs, which aren’t rare. For a variation to be considered a SNP, it must occur in at least 1% of the population. A significantly deleterious mutation – one that, decreases fitness by 1%, say – almost never becomes that common. Bad stuff has low population frequency. Yet the average person carries quite a few mutations with effect size of that order, each rare. You find them with highly accurate whole-genome sequencing, not SNP chips.

    So far, in the two papers that have covered this, Yorubans have about 25% more such than CEU or CHB.JPT (which have about the same), and black Americans have more than whites. Some kinds of mutations increase strongly with paternal age, some not: the ones that increase with high paternal age are the ones that are relatively high among the Yoruba. Who have a very high frequency of polygamy. Which leads to high average paternal age.

    I think that many researchers are at a disadvantage in this kind of research, since they either don’t know or actively deny the facts you might want to explain. Then there’s the fear of getting fired.

  3. a very knowing American says:

    “You can imagine two extremes. In one, everyone with schizophrenia has the same nasty mutation. In the other, virtually every family with members suffering from schizophrenia has a different causal mutation. That second pattern is bad news for medicine: it reduces the chance of finding a drug that helps a big fraction of sufferers. ”

    But even if the causal mutations are mostly different, they might mostly affect the same limited number of physiological pathways, so the chance of finding a helpful drug might not be so bad (depending on the details of the neurophysiology). This seems to be how height works: large numbers of genetic variants each having small effect, but mostly involved one way or another in skeletal growth.

  4. Pingback: linkfest – 12/02/12 « hbd* chick

  5. Imprint says:

    Greg, I am confused about your hypotheses about mutation load and IQ.

    “That Lohmueller paper looked at SNPs, which aren’t rare. For a variation to be considered a SNP, it must occur in at least 1% of the population. A significantly deleterious mutation – one that, decreases fitness by 1%, say – almost never becomes that common. Bad stuff has low population frequency. Yet the average person carries quite a few mutations with effect size of that order, each rare. You find them with highly accurate whole-genome sequencing, not SNP chips.”

    But on page 14 of this Beijing Genomics Institute presentation, they present an estimated distribution of minor allele frequencies for IQ affecting alleles (that are supposed to track half of IQ variation, although they mention that sub-10% frequency alleles may explain the difference between the 0.5 SNP estimates and some higher twin study estimates):

    https://www.cog-genomics.org/static/pdf/bga2012.pdf

    And they claim that these are mostly negative rather than positive alleles. But then you say:

    “What would a spelling-checked person, one with no genetic typos, be like? Since no such person has ever existed, we have to speculate. I figure that kind of guy would win the decathlon, steal your shirt and your girl – and you still couldn’t help liking him.”

    What portion of IQ variation do you think is attributable to alleles with frequency >0.1%, vs rare mutations with frequency <0.0001%?

    Also, for the spell-checked person, how do we know that the pieces would work well together? For example, if height and strength and such are all affected, and we have adapted to having a certain rate of mutational load, might the corrected version overshoot, e.g. by making us too tall for efficiency, or making some systems overdeveloped?

  6. Rob King says:

    Could it be more of an epigenetic effect linked to fast/slow life history? Cues to unstable environments might well push someone down the paranoid LHS route (we know it does on other measures) on the smoke-detector principal? E.g. its ok to be paranoid if enough of the environment really is out to get you?

  7. Pingback: A flood of new genetic variation - Evolving Economics

  8. statsquatch says:

    Wasn’t PCSK9 discovered by looking at rare gene variants in the Amish? That led to at least four phase III drug programs.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s