Sequences that are genes have function, but many other sequences are functional and are sometimes referred to as noncoding DNA. For instance chromosomes contains origins of replication, which define starting points of DNA replication. Also some parts of the DNA outside the genes are now known to be promoters of transcription or other regulatory sequences that figure in the control of gene activity .
"Junk" is somewhat a misnomer, as it is not yet clear whether these DNA sequences may turn out to be functional, but for better or worse, the term has stuck and is often found throughout the biological literature. As more functions are found the amount of DNA we now call junk DNA will likely decrease. It is also believed that junk DNA contains a pool of now defunct genes and gene fragments, known as pseudogenes, that were cast aside during evolution. Some hypothesize that the existence of junk DNA, which is not under natural selection, may provide a basis for further evolution, and thereby in a sense be functional. Some other evidence suggests that junk DNA is accumulations from failed viruses.
As much as 95% of the human genome is now considered junk DNA. Other species may have much more DNA than humans. For example the onion has 12 times as much, and it is believed that the excess is explained by junk DNA and pseudogenes. By contrast, the pufferfish has about 10 times less DNA, but has a comparable number of known genes. Therefore it seems that the ratio of functional and junk DNA differs widely per species.
It should be noted that junk DNA is a broader category than selfish DNA. The former is an expression of our ignorance, whereas the latter is a positive hypothesis which may or may not be descriptive of particular sub-sequences of a genome.
The 2003 November issue of Scientific American published an article "Gems in the Junk", in which it stated that "The failure to recognize the importance of introns 'may well go down as one of the biggest mistakes in the history of molecular biology'". [and] "What was damned as junk because if was not understood may, in fact, turn out to be the very basis of human complexity"
If this is true, science will not permit a "vacuum" for explanation(s) of how the former "junk DNA" (non-coding DNA) contributes to gene expression.
In an invited keynote address of the Electrochemical Society (Orlando, Nov. 15, 2003) the "FractoGene" explanation (and service) was introduced, that explains exons and introns as fractal geometrical sets (see abstract among "external links"). FractoGene attributes incremental improvements of gene expression, with more and more sophistication throughout evolution, and thus also explains why lower species have a smaller percentage of introns, while humans have the most, about 98.7%. Having sequenced both the mouse and the human, it is conspicuous that 99% of their genes are homologuous, and the major difference is that the mouse has about 1/3 less introns. (Fractogene.com)
http://www2.electrochem.org/cgi-bin/abs?mtg=204&abs=1195 FractoGene Design-Tool for Protein-Based Self-Assembling Nanostructures, Materials and Applications