It is my view that there is no useless DNA. That is, "junk DNA" is
a misnomer. There is DNA whose function is not understood. Several
years ago, i looked at the frequencies of adjacent bases in DNA of
different categories: e.g. tRNA, rRNA (coding, obviously), protein
coding, intron, and other (i.e. so called junk). This means that
i tallied up the frequencies of AA, AC, AG.... TT and compared them
to the frequencies expected given the frequencies of the individual
bases. You might expect if junk was junk that the frequencies are
rather unconstrained (doublet that is) and that they would differ
from those in the other categories. Not so. The frequencies in
different categories are well correlated. I suspect that junk
is involved in scaffolding, and other structural type things.
It may also (a la Cavalier-Smith) be involved in influencing the
amount of time taken for DNA replication, cell size etc.
A reference for the doublet work is:
Current Genetics (1988) 13: 523-530