Comments on: Will we ever… decipher everything about a life form based just on its DNA?

By: amphiox

amphiox — Thu, 08 Nov 2012 02:52:01 +0000

The answer is no. Because not all the information needed to define a lifeform is contained within its genome. The initial conditions in which the genome finds itself, the intracellular environment, also matters, and not all of that is set by the genome. Some of it is inherited as cells reproduce, soma to soma, in a continuous line all the way back to the very first cell.

By: Paul Knoepfler

Paul Knoepfler — Wed, 07 Nov 2012 22:41:59 +0000

What about the epigenome, which is basically the master of the genome? Cells literally do not know what to do with DNA without histones (which come in probably thousands of post-translationally modified forms that code for different DNA functions) and DNA methylation. The cellular reprogramming that gives us iPS cells is the best example of how the same DNA can manifest completely differently depending on the epigenetic context. So, a pile of raw DNA tells only a tiny bit of the story of any given organism.

Paul

By: James Sweet

James Sweet — Wed, 07 Nov 2012 21:31:05 +0000

I was going to say something similar to Nathan Meyers… Let’s say that practicality is an issue. Could you even in principle predict an animal from the DNA alone? Quite possibly not, because you have a literal chicken-and-egg problem: Even if you could simulate everything perfectly, you don’t know how the egg is going to develop unless you can simulate it being in a chicken. And if the only information you are starting with is the DNA, then you can’t simulate the environment of a chicken’s womb until you’ve already properly simulated the development of an egg.

MAYBE, with unlimited resources, you could reach some kind of convergence by performing countless simulations… egg in “generic” womb, then see what kind of womb that produces in the adult animal, then resimulate, lather-rinse-repeat until you start getting the same adult generation after generation. But even ignoring practical concerns, there’s no guarantee that the local minima you reach would be a “genuine” chicken!

tl;dr: I don’t think the DNA alone is sufficient information even in principle — never mind that, even if you could solve the “unknown womb” problem, the problem is so complex that it is likely to be intractable in practice.

By: Nathan Myers

Nathan Myers — Wed, 07 Nov 2012 11:24:36 +0000

So far as we have any experience, chicken DNA only results in a chicken if it’s grown in chicken cells. Similarly, elephants have only arisen from DNA that was contained entirely within elephant cells. Grow an elephant from DNA substituted into a chicken egg, and I will be the first to admit being impressed.

But let’s not jump the gun. Sure, an elephant cell can be hard to tell apart from a chicken cell, from the outside. It seems fair to guess that none of us have experienced the inside of a chicken cell, or an elephant cell either. They could be as different as a Honda Civic and a Mack truck, from that vantage point. The Mack truck is not going to be easy to find a parking place for downtown, and your “mobile” home is going to remain pretty stationary as long as it’s attached to your Civic, whatever its highway MPG rating.

What I’m really saying here is that chicken DNA has millions of years of practice in telling chicken cells what to do, but orchestrating the operations of an elephant cell is likely to stretch its skills beyond their limits. Sure, chicken and elephant cells both respire and metabolize. So do chickens and elephants, but that doesn’t mean a chicken can eat a whole tree, nor an elephant (Horton notwithstanding) hatch an egg. Can a chicken mitochondrion even use a protein transcribed from the sequences found in elephant DNA? I doubt it, and that’s just the beginning.

Nothing in nature makes a membrane from scratch; each bit of membrane that exists today started as part of another bit, going back to that first cell. Today’s cells can extend and pinch off new bits of membrane at a ferocious — indeed, given the resources, an exponential — rate, but if DNA can’t direct construction of a membrane from scratch, is DNA really so central to life? Really, DNA is just a tool to help the world-spanning membranium make more membrane. If a better way comes along, DNA will get the old heave-ho before you can say icosacatecholamine, and with no severance pay.

If you slurped out the whole innards of a chicken cell and, separately, an elephant cell, and swapped them, could you grow an elephant with chicken membranes or a chicken with elephant membranes? I’m far from sure that you could do even that. Would you grow the chicken in a culture medium tailored for chicken cells, or for elephant cells? I doubt they can be the same, but the cholesterol islands floating around in the envelope determine what gets in and out, at least at first.

Does a chicken Golgi apparatus process RNA precisely the same way as an elephant’s, or do they have slightly (or radically) different signaling and regulatory mechanisms? 300 million years is time for quite a lot of divergence. You might need to bring along a lot more than just the DNA to end up with a functioning cell. You might just need to bring along pretty much everything.

By: SteveW

SteveW — Wed, 07 Nov 2012 03:41:54 +0000

The problem with the physics analogy is that physics works best for predicting properties of a single entity: the location of a ball moving at a certain velocity, say, is governed by a single equation. If you introduce another object into the system, you can’t do much more than predict when or if they’ll collide. After that it’s all simulation, computing successive states of the system. So it is with DNA and cell reproduction. There is no single equation to predict the shape an organism will take after it divides more than once. The organism’s future states can be predicted to some extent using fractals and chaos theory, technologies that are computationally expensive and grow increasingly imprecise in proportion to the number of states.

By: Brian Too

Brian Too — Tue, 06 Nov 2012 23:50:05 +0000

In principle, yes.
In practice, this could take a couple of centuries.
Patience, Grasshopper.

By: Guy Plunkett III

Guy Plunkett III — Tue, 06 Nov 2012 22:38:40 +0000

On a lighter note, I am reminded of Dr. Margo Green’s Genetic Sequence Extrapolator “a computer program designed to describe the characteristics of a given species from a reading of its DNA … you can use this program to tell the species and sex of the animal, whether it was nocturnal, what it ate, how it hunted, how big it was…”

source: http://www.dailyscript.com/scripts/the-relic_early.html

By: DavidB

DavidB — Tue, 06 Nov 2012 19:26:30 +0000

Not sure I’d call the urethra a ‘stable environment’. Every so often…

By: Tim

Tim — Tue, 06 Nov 2012 17:04:15 +0000

Expanding on Hunter’s statements: If you take the view of computational biology, that the system of proteins, DNA and so on represents code and the processors which run that code, then you have a highly complex computer which not only reproduces itself (so that all life is, essentially, a Quine program), but actively reprograms itself. We already know from experience that self-modifying code is a debugging nightmare, and we know from theory and practice that many rather simple questions (presented in the form of algorithms) have uncomputable answers!

Whether life, or the weather, fits into this class is arguable I suppose — one could say that, since nature has obviously solved the equations, or ran the algorithms, up to this point in time, then an equivalent process can be computed in a purely abstract (mathematical) environment, also as a function of time. On the other hand, one could argue that the system, as a whole (for all time), is uncomputable, and life is merely the chaotic expression of an iterative algorithm flailing about, trying to converge on a solution that doesn’t exist. The practical question then becomes, “well I don’t care where life ultimately is going, I just want to know, what does this thing turn into after ten years?” And the answer is most likely: small increments are computable, so a finite time in the future is, in principle, computable; but even with really good computers, you may be waiting a very long time to get those results. (In other words… “an intractable problem.”)

By: Ian Sample

Ian Sample — Tue, 06 Nov 2012 16:46:54 +0000

Nicely done sir!
Your piece reminded me of a classic paper from Phil Anderson. Worth a read, still.

http://www.worldscientific.com/doi/abs/10.1142/9789812385123_others01