… but nothing at all to do with the book of that name by George Gurdjieff. Science Daily has an article about an interesting recent paper on genes and disease, that in effect sounds the death knoll for the genetic model of disease and opens a potential can of extremely hungry worms for biology as a whole.
Because the original article is behind a pay-wall, it’s a little hard to see what evidence the authors use to justify their new understanding, but by inference it would appear that their research into genetic markers of disease began to involve so many apparently irrelevant genes that a new approach seemed essential. Their new paradigm is this:
The gene activity of cells is so broadly networked that virtually any gene can influence disease, the researchers found. As a result, most of the heritability of diseases is due not to a handful of core genes, but to tiny contributions from vast numbers of peripheral genes that function outside disease pathways.
Any given trait, it seems, is not controlled by a small set of genes. Instead, nearly every gene in the genome influences everything about us. The effects may be tiny, but they add up.
The researchers call their provocative new understanding of disease genes an “omnigenic model” to indicate that almost any gene can influence diseases and other complex traits. In any cell, there might be 50 to 100 core genes with direct effects on a given trait, as well as easily another 10,000 peripheral genes that are expressed in the same cell with indirect effects on that trait, said Pritchard, who is also a Howard Hughes Medical Institute investigator.
Each of the peripheral genes has a small effect on the trait. But because those thousands of genes outnumber the core genes by orders of magnitude, most of the genetic variation related to diseases and other traits comes from the thousands of peripheral genes.
So, according to the old standard idea of “a gene for a disease” genetic defects cause diseases. But the Human Genome Project and its aftermath has more or less destroyed that as a myth – the anticipated medical fruits of that doctrine have proved to merely be a few shivelled prunes.
The next stage of the game was to suggest combinations of genes to be involved, the idea being that of “control genes”, “feedback loops” and other similar mechanisms drawing on machine metaphors. Like an electronic circuit, the job spec. here would be to untangle the loops and networks and draw, in effect, a circuit diagram for any given condition. Therapeutically you could then decide where you can break or augment the mechanism and change the course of the disease.
But this paper suggests, essentially, that every detectable thing in the genome affects everything else: to give that some perspective, note that the figure of 10,000 “peripheral” genes above is half of the total of protein coding genes in the human body; and given that not all those are expressed in every cell, the claim is that pretty well every gene in the relevant cells contributes to the disease. And you’ll also note that the cumulative effect of all these genes considered “unimportant” to the abnormal function outweighs the “core genes” to the extent of making them largely irrelevant to the outcome.
Now, diseases make good models for genetic studies not only because of the promise of medical benefits, but because they are abnormal states that, in principle, can be distinguished from normality and so more easily studied. But genetically there is nothing special about disease: it is detrimental function, in the sense that it is normality having problems – assuming (as we must) that there is a universal called “normality” from which diseases deviate.
In other words, if the “omnigenic model” is true of disease states, it is also true of every single normal function of the body. If virtually every gene is involved in contributing to, say, diabetes, then every gene will also be implicated in eyesight, flying, having rabbit ears or being able to solve quadratic equations in one’s head. If the “core genes” identified with heart function turn out to be “outvoted” by the mass-action of all the other genes, then a similar relativisation applies to genes hitherto thought to be associated with particular physiological functions, such as Hox genes and so on.
To put it the other way round, if pretty much every gene in the cell contributes to your being prone to breast cancer, then every gene in the cell is also a contributor to any specialised or generalised function you care to name. Think of any two functions: for example (out of thin air) digesting proteins and colour vision. Now pick any two ordinary genes with start and stop codons. The omnigenic model predicts that both genes will be involved, at some level, with both functions. Or instead make that 100 genes, or 10,000, and your two chosen functions – all those genes will be involved.
Alternatively keep your original two genes, and randomly select 100, or 10,000 functions or traits: those genes will be involved in all of them, or nearly all. Or that at least would be the norm, with the old idea that a gene codes for a protein, or a function, being at best an exception, or at worst an absurd over-simplification.
Near the end of the Science Daily piece is one of those “what does it all mean?” conclusions:
Pritchard’s omnigenic model promises to take basic biology in new directions and means biologists need to think a lot more about the structure of networks that link together those thousands of peripheral disease genes.
“Encouraging new lines of research have opened up” is the gist of that. But think about it: what does it really mean to explore the “structures of networks that link together those thousands of peripheral disease genes”, when all they are in reality is exactly the same “peripheral genes” that make us, jellyfish or sequoias what they are? What kind of circuit diagram can one adduce from knowing that that the entire genome is responsible for every individual thing that genes do in the organism? Surely the entire model is as unworkable as suggesting that all we need to do to understand human society is to understand all the links between each human being. It may be true, but one needs to be God to do it.
Perhaps – perhaps – such complexities may be understood in principal, but not, surely, by reductionist sciences and their simplistic “control networks” modelled on computers or heating systems. One has instead to think in terms of totalities, and their overall teleological aims – not about which bit controls what.
A medical implication of this “every gene counts” concept of disease is that it pushes one back towards the Hippocratic model and away from the reductionist one we have adopted. The latter is about removing specific causes of illness (such as engineering genes, replacing your mitochondria and similar drastic and expensive modifications), but the former is about restoring the balance of the body itself. That appears to make more intuitive sense if your disease is actually caused by your whole genome (and maybe even more than just the genome) .
But if that non-reductionism is true when we consider the understanding of biology as it functions moment by moment, it’s also as much a challenge to existing theories of evolution. Forget the idea of tracking the fate of individual genes: what’s the point when there’s no definite relationship between that single gene and anything happening in the living organism, which (according to the new model) depends on all the genes at once?
But does it not also make the idea of evolution by variation and natural selection meaningless, if we understand that in terms of some trait being advantageous or not, surviving and being retained? Adaptive natural selection has long been subject to the criticism that it will rapidly be swamped by the competing selectivity of multiple traits. That is a major plank of Kimura’s neutral theory.
But how much more is natural selection unintelligible when none of those traits is separable by cause and effect! Being taller might confer a selective advantage. But if, as Pritchard says, that tallness arises from the interaction of many thousands of varying genes, whose variations also affect each and every other trait (and disease) in the organism – traits having every possible variety of advantage, disadvantage or neutrality – then how could selection of a particular trait ever be targeted by the environment?
Stephen Jay Gould pointed out how adaptationism is crippled by the idea of a “spandrel” – which he conceived as a genetic trait getting a free ride with another, adaptively advantageous, trait. But in omnigenism, any selected trait carries many thousands of other traits along with it. You think somebody will make a difference to your company’s success, but you find he won’t work for you unless you employ his entire extended family of shirkers and disabled.
To me there is one obvious conclusion, if (as seems probable from the way genomics is moving) some such picture as omnigenism obtains in life. You have to begin to think of both organismal function and evolution in terms of some global principle of direction either from within the organism or from outside it: that is, some principle of teleology. The first of those seems remarkably like the long-discarded vitalism. The second appears indistinguishable from special divine providence.