Junk or No Junk?

The phrase "junk DNA" is a hot button. Authors of press releases, news stories, and even some journal articles seem powerless to resist casting any new discovery of function in non-protein-coding DNA as overthrowing a cherished belief that most of the DNA is junk.

In contrast, bloggers including T. Ryan Gregory and Larry Moran regularly gripe that this framing, like many "people used to think x, but now…" stories, is misleading: biologists have known for decades that non-coding DNA contains important regulatory and other functional sequences. Nobody seriously thought it was all junk: that's just a myth that makes the story seem more exciting.

Still, most biologists agree that DNA is mostly junk.

John Mattick is not so sure.

In a 2007 paper in Genome Research, Mattick and his co-author Michael Pheasant, both of the University of Queensland in Australia, suggested that evolution could be sheltering much more than the 5% of the genome estimated by the ENCODE project and others. Those researchers estimate the background rate of neutral evolution by looking at sequences that they assume to be useless, such as "ancient repeats" left behind from long-ago genomic invasions.

Instead, if these sequences are a little bit useful, perhaps because they are occasionally drafted by the cell for other uses, then they would be slightly preserved during evolution. If this is the case, then other sequences that are also slightly preserved may be useful, too.

I discussed this issue with Mattick for my story for Science (subscribers only), but it was hard enough to capture the issues for strongly selected sequences, so this subject didn't make the cut. Mattick didn't claim that the issue is settled, only that the logic was in danger of being circular: "It is basically an open question. We have no good idea how much of the genome is conserved, except for that which is dependent on questionable assumptions about the nonconservation of reference sequence."

For him, the extensive transcription of the genome seen by ENCODE may not be a sign that RNA production is unselective, but that a large fraction of the DNA is serving some useful, if so far unknown, function.

Mattick described himself as a "minor author" among the scores on the ENCODE project. Ewan Birney, of the European Bioinformatics Institute in the U.K., played a coordinating role. But he doesn't strongly dispute Mattick's observations. "Ancient repeats provide a marker of evolution. They may very well be under some selection," Birney said. But "the striking thing," he stressed, "regardless about where the line is between selection and no selection, is that a lot of the functional regions are at the absolute lowest end of what we see across the human genome." They may be important, but not very important.

Or maybe the biochemical assays don't measure biological importance at all. "Many people instinctively feel," Birney said, "that all the functional elements really must be selected for in some sense. But there's alternative view, which is that there's just a big set of cases which are generated randomly, are perfectly assayable, when you assay them they're always there in that species, but in fact evolution doesn't select either for or against them. They are truly neutral elements: they are selected neither for nor against."

More philosophically, Birney doesn't find the evolutionary question to be central to short-term concerns about human health. "For disease biology we're interested in understanding the disease. We're not so interested in proving whether they're weakly under selection or something like that."

In fact, for both Birney and Mattick, the extensive biochemical activity of the weakly selected 95% of the DNA suggests its potential as a reservoir of "spare parts." Whether or not the long-term potential of that reservoir puts evolutionary pressure on its components, so that they are conserved, may not be the key issue. The important thing is that those partially-assembled genetic tools are ready to be called into action for future innovations.