GeNets

Arguments supporting the view that, the portion of DNA which used to be considered "junk" is actually where the most important [embryo] development and other important biological processes are encoded, have been put forth by the molecular geneticist Prof. John S. Mattick of Queensland, AU. Mattick's arguments are lucidly presented in: "Non-coding RNAs: the architects of eukaryotic complexity".

Mattick's observations are:

The "genes", i.e. the DNA segments that code for the proteins account for only 1.5% to 2% of the human genome and similarly low percentages of the genome of other species sequenced up to now.
The amount of DNA code in the remaining "junk" DNA increases as one goes up the scale of species complexity.
Sequencing results show that human and chimps share better than 98% of genes (protein coding segments), similarly humans and mice share better that 90% of such genes. This suggests that the obviously large differences that exist between such species would be specified by the better than 90% of DNA sequences not involved in coding for proteins.
The eukaryotes' DNA differs from the prokaryotes' DNA in that the former protein coding sequences are chopped up into protein coding segments: exons and non protein coding segments: introns. The latter are so named because they are intragenic sequences. He suggests that this fundamental change arose because the prokaryote genetic architecture would be limited in expressing organism complexity.
He further suggests that the introns and other non coding segments of eukaryote DNA are the locus of signaling regulatory networks capable of orchestrating the expression of numerous and complex networks of "genes".
He suggests that such signaling networks could be similar to computing and dataflow networks.

These arguments dovetail very well with the recent findings from the evolutionary embryologists (Evo Devo) of master control sequences which seem to trigger cascades of protein based signals turning on and off controlled genes. With signaling proteins binding to code segments (signatures) of the controlled genes the latter acting in the manner of logical networks. All is needed is that some or all of the introns be the locus for the Evo Devo "signatures"

The Evo Devo signaling networks are potentially multilayer nets. With a Master Control Gene (MCG) controlling a first layer of genes, which, in turn, can control another layer of genes, the second layer in turn controlling a third and so to any required depth.

From a computational point of view, the so called "junk" DNA has to include, not just the generation of a large array of RNA signaling molecules, but also some way of generating memory for such biological computations. In fact, logic networks with no memory or state can not implement sequential processes or computations. An obvious candidate for the computation state is the actual collection of cells & cell types present at a particular time.

What is kind of strange, at this point (2006) is that the molecular geneticists seem to either not know or ignore what the Evo Devo people are doing. No matter how I searched I could not find on the Web any papers linking the two kinds of investigations. If this disconnect is real it could be explained by the fact that the two communities are working at very different levels of the overall biological architecture. With the molecular geneticist working directly with the hard evidence from DNA sequences. This is like trying to infer the role & semantics of a complex program such as an OS by reading just sequences of binary code.

The Evo Devo community, on the other hand, is working with the evidence arising from either altering DNA or splicing DNA segments from one species to another and observing the effects on the development of the eggs so modified. The focus being on MCGs responsible for the development of major organs or parts of the organisms. The Evo Devo community is therefore, working at the much higher, and more abstract, level of a species BioPert. Ultimately these two viewpoints will have to converge on a grand synthesis including both of them, but of course we are, at this point, very far from accomplishing such a synthesis.

It is clear, from an information processing point of view, that we must focus on the 90% of the DNA which instead of junk is indeed where most of the biological program is encoded. In the parent page I summarize what is known from embryology. In fact embryo development is the "output" of the biological program, and its study, especially across species, will suggest: its components, timing of its sequences and spatial rules for its operation.

Finally, the fact that the eukaryote branch of life may have adopted a significantly diverse and more expressive DNA molecular architecture may explain the Cambrian explosion. In fact, if a new class of organismic architectures became feasible because of the adoption of much more expressive development control mechanisms, it would be quite understandable that a large number of new embryogenesis programs would evolve, in an relatively short 40 MY.