Study turns human genetics on its head

November 23, 2006 2:37 PM

Study turns human genetics on its head. "The genome is like an accordion that can stretch or shrink . . . so you have no idea what's normal. We have to think of genetics in an entirely different way. We're actually more like a patchwork of genetic code than bar codes that line up evenly. Everything we've been taught is different now."
posted by ZenMasterThis (28 comments total)

Here I thought this was going to be a discussion about epigenetics. Turning cancer off, cultural memory written into our biology? This is the cool stuff. (20 minute mp3 behind that link, I highly recommend it, totally fascinating stuff.)
posted by Hildegarde at 3:00 PM on November 23, 2006

This nature paper is interesting, but there's also a good amount of self-promoting hyperbole on the part of the researchers. It doesn't help that the article's slant can be summarised as "All the (not quite correct) assumptions I had about genomics might be wrong!"
posted by docgonzo at 3:14 PM on November 23, 2006

Free full-text study is here
posted by rxrfrx at 3:21 PM on November 23, 2006

I knew it!
posted by jimmythefish at 3:22 PM on November 23, 2006

Short version: Although we're all 99.9% the same in terms of what sequences we possess, we're way more different than previously though in terms of how many copies of these sequences we've got. Copy number likely affects phenotype in most instances where the gene codes for something important.
posted by rxrfrx at 3:24 PM on November 23, 2006

I dunno, I first saw this headline on Drudge, which means it's automatically not true, right?
posted by interrobang at 3:25 PM on November 23, 2006

researchers have found that while people do indeed inherit one chromosome from each parent, they do not necessarily inherit one gene from mom and another from dad.

I'm guessing that's not actually what they found, because if it was it would have been nothing new. We've known for a long time that people (and all animals) can carry multiple copies of various genes. In some cases, those multiple copies then mutate to create new things without destroying the original (sometimes necessary) gene. I wonder what it is they found that got them so excited, If anyone wants to pay $18 they can read it.

According to the editor's summary it's not a new discovery at all, but simply a more complete mapping of the differences in people's genetics. In other words, very useful and important information, the result of a lot of work, and something that would be very intresting to geneticists, but it's not at all like what the globe and mail author (Carolyn Abraham) makes it out to be.
posted by delmoi at 4:55 PM on November 23, 2006

Free full-text study is here

That's just the abstract.
posted by delmoi at 4:57 PM on November 23, 2006

Gene duplication has long been known to exist and is the necessary source of gene families - duplicate genes that evolved to take on slightly different functions.

In most I've read, duplication (of the whole genome or just one gene) is considered an accident that happens.

This study seems to say there may be some logic behind duplication.
posted by MonkeySaltedNuts at 5:04 PM on November 23, 2006

#delmoi:
That's just the abstract.

Um, from that page you can click on several versions of the full-text study.
posted by MonkeySaltedNuts at 5:05 PM on November 23, 2006

Just had a look at the paper, but rxrfrx pretty much summed it up. We're more variable than we thought we were.

I think the real implications of this work will be in regard to genetic associations for disease. Although variable copy number has been associated with a number of disorders (as written in the introduction), there's a lot that hasn't been explored.

And MSN, while everything you said about gene duplication is true, duplications of gene families are usually talked about in terms of comparisons between different species, not comparisons between different individuals within the same species.

The emphasis isn't on something new, so much as something that is far more widespread than previously thought.

Thanks ZMT!
posted by kisch mokusch at 5:36 PM on November 23, 2006

Free full-text study is here

That's just the abstract.

There's a centerfold. Or accordion something. It expands out. Nevermind.
posted by hal9k at 5:39 PM on November 23, 2006

Um, from that page you can click on several versions of the full-text study.

Oops, you're right. Actually it looks like I linked to the wrong article above, it was a different article about gene duplication in this month's Nature

This study seems to say there may be some logic behind duplication.

Since we've found the full text of the article, can you cite some textual basis for that claim? From what I'm reading, the major claim is that most phenotypic differentiation in humans is caused by CNV (copy number variation), rather then actual mutation in specific variations.
posted by delmoi at 5:46 PM on November 23, 2006

Another Short Version: When they said you had your dad's eyes and you mom's nose, it turns out that was true.

I have a brother. He's shorter than me. He's balding and I'm not. This is all despite the fact that our parents' genetic make up did not change between the time he was born and I was born (or at least mutated only slightly). So my brother, my closest genetic relative, is very different in appearance and body structure. Why should it shock me that people from another ethnic group should have a different genetic make up. I don't understand how, "everything we've been taught is different now," when I can simply go hang out with my brother and see this to be true?
posted by Pollomacho at 6:21 PM on November 23, 2006

#kisch mokusch: And MSN, while everything you said about gene duplication is true, duplications of gene families are usually talked about in terms of comparisons between different species, not comparisons between different individuals within the same species.

conventional discussion doesn't even have to involve comparing species. That a gene family exists implies there was a duplication, and you might be able to use neutral substitutions to get some idea of its age. However if you do compare compare species, such as the human and sea urchin genome, you can tell which gene families were present in the common ancestor, which genes (or whole genomes) were duplicated since then and sometimes even if genes were duplicated and lost.

But this family perspective on duplicate genes was that a whole species some how wound up with two copies of a gene. Not (as the present study shows) that the number of gene copies can (?for some genes) vary widely within members of the same species.

It is less necessary to say that a 'gene duplication' event took place. CNV seems so widespread that duplication is a general mechanism, not some special event.
posted by MonkeySaltedNuts at 6:25 PM on November 23, 2006

I have a brother. He's shorter than me. He's balding and I'm not. This is all despite the fact that our parents' genetic make up did not change between the time he was born and I was born (or at least mutated only slightly). So my brother, my closest genetic relative, is very different in appearance and body structure.

Well, this is possible under the 'old' Mendelian model of dominant and recessive genes, As well as the 'on it's head' model described here.
posted by delmoi at 7:35 PM on November 23, 2006

"Here I thought this was going to be a discussion about epigenetics. Turning cancer off, cultural memory written into our biology"

Hildegarde: I don't mean to single you out (and I feel bad because I didn't listen to the mp3 before writing this and it may already say all of this and maybe I'm misreading what you were saying) but please please please, let's not go around talking about epigenetics as a way of preserving cultural memories or anything like that. 'Epigenetics' simply refers to changes in levels of gene expression that are relatively stable over the life of either an organism or, more typically, a tissue.

All this means in a practical sense is that cells have the ability to permanently turn off some genes while leaving others on, even after indefinite rounds of cell divisions. A neuron is not a hepatocyte because, among other reasons, it has turned off expression of DNA pertaining to hepatocyte-specific genes --- these genes are still there, they just aren't expressed. By the same token, neuron-specific genes aren't present solely in neurons, but they are expressed at higher rates in these cells. The point is, there is no change to the genome. If that neuron wanted to go and turn on those hepatocyte genes, they're still there, just waiting to be opened up. Nothing is gained, nothing is lost.

There is absolutely no known or forseeable genetic basis for inherited memories --- a predisposition towards certain behaviours or modes of cultural expression, sure, absolutely --- but bonafide inherited cultural memories? Not a chance.

I feel strongly about this because it seems like so many of these amazing breakthroughs in biology get blown out of proportion by the media and the public, until what was once an elegant piece of scientific enlightenment becomes a little CSI factoid that kind-of-sort-of-not-actually gives creedence to some old myth. Like one time I was watching one of these 'HIV Doesn't Cause AIDS', um, for lack of a better word 'documentaries', and I just wanted to throttle everybody who had made it; these people knew just enough about immunology and molecular biology that it was apparent they should have known better. They got the terms and most of the core ideas right, but managed to mangle pretty much every other detail while burying their ignorance in a pile of flimsy metaphors. I looked around at the non-biologists in the room, though, and they all seemed to be going 'yeah, good point.' A little knowledge is a dangerous thing, indeed.

/derail
posted by Tiresias at 7:44 PM on November 23, 2006 [3 favorites]

My point exactly. I remember being taught about such-and-such gene duplication event, which took place so many millions of years ago. Clearly gene duplication is not as uncommon as previously believed.
posted by kisch mokusch at 7:44 PM on November 23, 2006

kisch mokusch has it. Before this paper, no one disputed that gene duplication and deletion events happen. It's just that it was assumed to be a rare, often pathogenic, event -- see, for example, Scherer and Osborne's earlier paper on a duplication that appears to cause expressive language delay. Interestingly enough, when the same region is deleted, i.e. there's only one copy present in the person's genome, the result is Williams-Beuren Syndrome, in which expressive language is actually enhanced. The new data show that copy number variation is a _lot_ more common than we thought it was.

/disclaimer: I work in Dr. Scherer's lab, albeit on things that have nothing to do with this particular discovery.
posted by greatgefilte at 9:26 PM on November 23, 2006

Wow, This is a pretty cool paper. There are heaps of interesting bits in it, but two jump out:

1. All those complex disease related SNP association studies are going to have to be reevaluated with this data in light. Most of the tests used assumed this kind of stuff didn't happen, and it could lead to a lot of old papers being bunk.

2. These CNVs must be kind of new, judging by figure 7, most of the variation has happened since the split between Europeans, Asians and Africans (as an aside, that is the clearest separation of races I've seen by genetic data, look for someone in the press to notice and shit to hit the fan soon). It seems that the dismissal of the 0.2% of new CNVs they found in this generation as errors might need to be revised.
posted by scodger at 1:40 AM on November 24, 2006

"as an aside, that is the clearest separation of races I've seen by genetic data"

But population genetic studies have always shown ethnic groups, and if you scale back enough you'll see European, Asian, and African groupings. But race isn't a concept rooted in known relatedness and then extrapolated into a larger racial grouping, it's a concept rooted in taking a cluster of superficial characteristics and then assuming relatedness. Some of those traits correlate well to relatedness in many or most contexts, others correlate well in limited contexts but not in greater contexts, and still others don't correlate very well to relatedness very much at all. But our modern conception of "race" ignores where those shared traits don't indicate relatedness because a) we have a strong instinct to generalize based upon physiological superficialities, and b) people today actually believe a sort of folk genetic racial theory that these superficial traits are very reliable markers for relatedness.

So what does that mean in the context of this study? Well, in light of this study we'll have to consider the possibility that genetic differences between subpopulations can be larger than we thought they could. But it still doesn't validate the modern idea of race because, simply put, the modern idea of race asserts that all black Africans are a "race" and are more closely related to each other than they are to any non-black skinned population. Which we know isn't true and that hasn't changed.

For the modern notion of "race" to become scientifically accurate, it will have to accommodate the reality that its current predominant identifying trait—skin color—is not a reliable indicator of relatedness. In the end, it would have to become something that simply matches population genetics studies of relatedness which simply would not allow the gross groupings that the current notion of "race" relies upon. And a corrected notion would largely correspond to ethnic groupings, which we already acknowledge anyway. And in any event, an adjustment to the scientific reality of the situation isn't going to happen because at rock bottom the notion of race is about distinguishing between white and black people and claiming they are distinct in a fundamental sense.
posted by Ethereal Bligh at 4:47 AM on November 24, 2006

This is a whopper. This is big.

Chromosomes were like totem-poles, an endless stack of concretely positioned but paired and perhaps contrasting genes.

If it holds, this CNV screws a lot up. Try teaching Mitosis or Meiosis then, good luck (if you've got good students). DNA replication? What--do DNA polymerases feel nostalgic and bail a ssSequence in order to back-flip and recapitulate?

Watson and Crick and Franklin saw, in this maverick structure, the potential for faithful replication. This thing flies in the face of that assumption.

The coarse material for rapid evolution scores again.
posted by toma at 5:10 AM on November 24, 2006

I, for one, welcome back our old genetic overlords.

The hopeful monster thing was cool. I'm one myself.
posted by ewkpates at 9:44 AM on November 24, 2006

If it holds, this CNV screws a lot up. Try teaching Mitosis or Meiosis then, good luck (if you've got good students).

I don't see how this affects mitosis or meiosis.

DNA replication? What--do DNA polymerases feel nostalgic and bail a ssSequence in order to back-flip and recapitulate?... Watson and Crick and Franklin saw, in this maverick structure, the potential for faithful replication. This thing flies in the face of that assumption.

Don't overstate this. It's already well-known that these sorts of duplications exist, so there's no new molecular biology here. No one assumes that DNA is reproduced without error. If that were true, then there would be little evolution.

What is new is the realization that copy number variation is much more common than we thought in humans, and potentially more important than other kinds of variation.

For a long time we have known that expression levels are an important cause of inherited variation, and assumed that these are caused by polymorphisms in DNA sequence outside the protein-coding regions (what some people once referred to as "junk DNA"). There is a subtle and not yet understood genetic code that relates these changes to changes in expression, and a lot of us are very interested in this area. To find out that a lot of the difference is due to the number of gene copies would be somewhat disappointing.
posted by grouse at 9:56 AM on November 24, 2006

Also, Ethereal Bligh has it exactly right.
posted by grouse at 10:05 AM on November 24, 2006

Don't overstate this. It's already well-known that these sorts of duplications exist, so there's no new molecular biology here. No one assumes that DNA is reproduced without error. If that were true, then there would be little evolution.

What was accepted was low-rate replicative errors that become the fuel for genetic variation. And gene duplication, translocations and deletions were interesting, but rare. Taught as "Hey, look what diseases can be produced by these mistakes."

If this is true, it supports the notion that whole chunks of DNA are routinely copied or deleted, probably as an evolutionary strategy. The whole process of DNA replication can't then be as faithful as it once appeared, hence my bringing up Meiosis and Mitosis. If it's so prevalent that it can screw up an amniocentesis screening, it can't be trivial.

And, strictly speaking, it might not be new molecular biology, but it's certainly a new understanding of it. The 'blueprint' had always been sacred, deviations were dangerous but occasionally rewarding, given the long arc of evolutionary competition. Now it might be that DNA inheritance is more of a gamble than we thought.
posted by toma at 4:04 PM on November 24, 2006

No one assumes that DNA is reproduced without error. If that were true, then there would be little evolution.

It is just that the old overriding view of reproduction error is "single-nucleotide substitution" (see Single nucleotide polymorphism^WIKI).

From a SNS perspective one could view gene-deletion as a SNS that makes a gene not work (either because the slightly different protein doesn't function correctly or that a SNS puts a premature STOP in the middle of a protein code). Evolution eventually throws the defective gene out of the genome.

In the new perspective SNS reproduction errors are reduced to more or less equal footing with whole gene duplication and removal.

The big question is how much "more or less" this is.
posted by MonkeySaltedNuts at 4:43 PM on November 24, 2006

Wow, a bunch of geneticists decide to PR the hell out of their paper and look what happens.

OK, so they found 1400 CNV regions covering 12% of the human genome, using whole-genome SNP analysis. Of these regions, 30% were previously known. Regular genes are underrepresented, while CNS is overrepresented in that 12%. Cell adhesion proteins are highly overrepresented, which I think means most protein-coding CNV diversity is immune system related. They can't map the exact copy number at each locus, the positions of indels/duplications, or the integrity of the genes in those regions, which limits us to speculation as to how exactly this might affect expression. If I understand correctly, for all we know, 90% of these CNVs could be accounted for by Alu proliferation with no expression impact; alternatively, there really could be 3000 genes and a ton of CNS whose expression is variable by copy number.

This clearly highlights the fact that we should pay a lot more attention to CNVs (which to me professionally means I better start paying attention to all that SNP data). This is pretty exciting, but the sky is not falling on genomics.

Also, toma, could you please refrain from uninformed sensationalist statements? You're like that documentary Tiresias mentioned: it's obvious that you know just enough to know better.
posted by azazello at 9:01 PM on November 26, 2006