Some does, some has no apparent purpose at all, and is very unlikely to. (various repeating sections, pseudogenes, fossil viruses to name a few.)
Remember the LGGLO (scurvy) mutation in the great apes (including people). It's very similar to functional dna in other mammals, except for one missing base pair, which totally screws it up downstream from there.
Is it there for some unknown regulatory purpose? Evidence? Is it there to make a novel protein that only apes need? Evidence?
If it's not junk (your claim) and it would allow us to make vitamin C if just one base-pair were added (sequencing data), then the conclusion would be that it's a mutation that does something useful. But you're always claiming that doesn't happen!
Some does, some has no apparent purpose at all, and is very unlikely to. (various repeating sections, pseudogenes, fossil viruses to name a few.)
One of those 'highly repeating sections' was the ALU sequences in over 10% of the human genome. They are needed for cell reproduction. They act as a zipper when a cell divides in two. That is why it is very dangerous to call anything we do not know 'useless'. The discovery of the control systems of genes is very difficult. The control regions can be very large as the following shows:
Correction of homozygous and heterozygous Small eye phenotypes has been achieved with a 420 kb human YAC containing the 25 kb PAX6 gene, and about 200 kb genomic flanking sequences on each side. The same YAC, truncated 7-10 kb downstream of the intact PAX6 gene failed to correct the homozygote lethality or eyeless phenotype, although the heterozygote phenotype was ameliorated (Schedl, 1996). We have generated transgenic mice with a shorter YAC, extending approximately 120 kb downstream of PAX6, approximately to the most distal aniridia breakpoint, and thereby about 80 kb shorter than the YAC which was shown to rescue the Smalleye phenotype. We show that this shorter YAC is unable to rescue the Smalleye phenotype in those transgenic mice, suggesting the presence of a regulatory control region more than 150 kb downstream of the PAX6 promoter(s), in the 80 kb region between the shorter and the longer YACs.
We have carried out a DNaseI hypersensitive sites analysis of that region using PAX6 expressing cell lines and a non-expressing cell line and have identified 8 hypersensitive sites spread out over a 25 kb region. We show that a fragment carrying a subset of these hypersensitive sites is able to direct the expression of a LacZ reporter gene in the eye and nasal epithelium of transgenic embryos. Through interspecies comparison we have identified another, conserved DNA element in this region, which does not form a hypersensitive site in the used cell lines, but directs LacZ expression in the lens, fore- and hindbrain of transgenic embryos. Taken together these results provide evidence for the presence of a complex regulatory control region at a large distance downstream of the PAX6 gene.
From: Analysis of Regulation in Pax-6
In English, what the above means is that 'there is a lot of regulating going on' outside of the gene itself and it is very hard to find exactly what does what.