The "best possible code"

A paper written by Knight et al in 2000 has created some confusion as to the nature of the genetic code, leading some design proponents to jump to the conclusion that these findings show evidence of ‘design’ when in fact, the findings, in proper context show strong support for an evolutionary thesis of the origin and evolution of the genetic code. Let me explain.

The paper in question is:

Early fixation of an optimal genetic code published in Molecular Biology and Evolution 17:511-518 (2000) and written by Knight, Freeland, Landweber and Hurst.

The quote that caused much confusion is

If our definition of biosynthetic restrictions are a good approximation of the possible variation from which the canonical code emerged, then it appears at or very close to a global optimum for error minimization: the best of all possible codes.

which was quoted by ChunkDZ as (warning, many of ChunkDZ’s responses include insults, invectives, follow the links at your own risk).

ChunkDZ wrote:

If our definition of biosynthetic restrictions are a good approximation of the possible variation from which the canonical code emerged, then it appears at or very close to a global optimum for error minimization: the best of all possible codes.

Later abbreviated to (warning, many of ChunkDZ’s responses include insults, invectives, follow the links at your own risk).

ChunkDZ wrote:

After all your lying, posturing, obfuscating, pretending, and doubletalk, several scientific discoveries remain unchanged.

1) The code is “at or very close to a global optimum for error minimization across plausible parameter space.”

2) The code is “the best of all possible codes”.

However, the abstract itself could have been a hint as to the nature of the claim

Finally, other analyses have shown that significantly better code structures are possible. Here, we show that if theoretically possible code structures are limited to reflect plausible biological constraints, and amino acid similarity is quantified using empirical data of substitution frequencies, the canonical code is at or very close to a global optimum for error minimization across plausible parameter space.

or from the actual paper

Estimates based on PAM data for the restricted set of codes indicate that the canonical code achieves between 96% and 100% optimization relative to the best possible code configuration (fig. 2c).

So what went wrong in ChunkDZ’s analysis of the quote? Note that ChunkDZ argued that the paper showed that the code was globally optimal, which indeed could be seen as evidence against evolution and in favor of design since evolution seldomly optimizes globally but rather finds a local optimum given the initial conditions which are guided by historical contingencies and other constraints. However, a ‘designer’ has no limitations to how the optimization is performed and thus could, if it pleased her, optimize in a truly global fashion. When you read the paper, the authors also looked at the unrestricted code performance and as expected, the standard code fared much worse placing its performance at between “76% and 97%” of optimal value.

ChunkDZ realizes his problems when he states

ChunkDZ wrote:

It’s globally optimal within all plausible biological constraints.

but fails to recognize that these are not biological constraints, since any code could function as the genetic code. However it is the historical constraint based on the prevailing hypothesis of the origin and evolution of the genetic code which proposes that pre-biotic chemistry determined the link between the code and the amino acid assignment (this concept is called stereochemistry) and that subsequent optimization further shaped the code to some ‘optimum’ (also called selection). Under this hypothesis, the analysis shows that indeed, given the hypothetical pre-biotic linkage, the code is indeed optimal in the sense that no or few codes which similarly reflect pre-biotic conditions exist which do better (the 96 to 100% claim). In other words, the code is only globally optimal in a constrained fashion, as expected from evolutionary theory, and not as expected from a ‘design’ perspective unless one restricts the designer to be constrained by the similar pre-biotic chemistry. In fact, the global nature of the optimization was an essential argument in ChunkDZ’s claim that the research supported ‘design’ and now that it has been shown that in fact the opposite is true, one comes to understand how ChunkDZ may have missed the important limitation of the claim.

ChunkDZ’s claim is further clarified by the following comment (warning, many of ChunkDZ’s responses include insults, invectives, follow the links at your own risk).

ChunkDZ wrote:

My point was, and remains, that you moron critics love to point out that nature is expected to make kludgy hodge-podges. Then when confronted with evidence that the basis for every single biological system is a non-hodge-podged, sophisticated, elegantly designed, optimal “best of all possible codes”, you monkeys simply wallow in your own feces, ignore the research, and complain about the evils of Big Bad Billy Dembski.

What a predictable bunch of pansies you are.

Anyhow, here’s another research paper that shows that the code is also optimal for parallel coding.

http://genome.cshlp.org/cgi/content[…]act/17/4/405

Note the ‘best of all codes’, as opposed to a historically constrained optimum, which suggests to ChunkDZ not only that evolutionary theory seems flawed, even though it was an evolutionary prediction that suggested that the code originated from a linkage with pre-biotic chemistry and was then optimized ‘locally’ and not the best of all codes in the sense of an unconstrained optimum, which as I explained would have been a problem for evolution and add some relevance to the concept of ‘design’, assuming that ‘designers’ are not constrained by pre-biotic chemistry and are interested more in a truly global optimum. In fact, and this is worth repeating, there are no biological reasons why the code has to reflect the pre-biotic chemistry, in fact, the code could very well have been one which was totally unrelated to said chemistry and still function biologically.

Note also how ChunkDZ in the above quote refers to another paper which he believes claims ‘optimality’. Let me quote from the actual title of the paper “The genetic code is nearly optimal for allowing additional information within protein-coding sequences”. Note the word nearly, in front of ‘optimal’. Furthermore, the conclusions again do not show a global but rather local optimum, but it does require one to read the paper.

And finally, the coup de grace, so to speak. Knight was one of the reviewers of a more recent paper which looked at the global optimality of the genetic code, where the code was obviously not constrained by pre-biotic chemistry to show that the code was far from a truly global optimum, and Knight observed that

They recapture the uncontroversial result that the genetic code is much better at minimizing errors than a random genetic code (as has been shown by many authors), but is at neither a local nor global optimum (as has also been shown previously).

What more needs to be said but… ‘priceless’

Except of course, ChunkDZ seems to be unconstrained and argued that

Compare the design hypothesis according to ChunkDZ

the designers (far from being constrained) chose the ones that they used for their own reasons.

versus the scientific hypothesis

The origin of the genetic code was constrained by pre-biotic chemistry (stereochemistry hypothesis) followed by a period of selection

Now which one do you think is the better hypothesis and why?

Oh and about this more recent paper?

Artem S Novozhilov, Yuri I Wolf, and Eugene V Koonin Evolution of the genetic code: partial optimization of a random code for robustness to translation error in a rugged fitness landscape, Biol Direct. 2007; 2: 24.

Instead of analyzing just the restricted code set, they analyzed the full code set and found some interesting results

  1. The code fitness landscape is extremely rugged such that almost any random initial point (code) tends to its own local optimum (fitness peak).

2. The standard genetic code shows a level of optimization for robustness to errors of translation that can be achieved easily and exceeded by minimization procedure starting from almost any random code.

  1. On average, optimization of random codes yielded evolutionary trajectories that converged at the same level of robustness as the optimization path of the standard code; however, the standard code required considerably fewer steps to reach that level than an average random code.

  2. When evolutionary trajectories start from random codes whose fitness is comparable to the fitness of the standard code, they typically reach much higher level of optimization than that achieved by optimization of the standard code as an initial condition, and the same holds true for the minimization percentage. Thus, the standard code is much closer to its local minimum (fitness peak) than most of the random codes with similar levels of robustness (Fig. 9).

  1. is the clincher which shows that a genetic code from almost any initial condition could evolve to reach an optimum as good or better than the standard code.

So much for ‘best of all possible codes’

Now you know “the rest of the story”, and you have been reminded once again why ID is a scientifically vacuous position.