Little is known about the development and evolution of brain features that are specific to humans, and work in this area has so far focused mainly on protein-coding regions of the human genome. But more than 98% of the human genome is made up of non-coding DNA. So, at the start of a search for evolutionarily important non-coding stretches, David Haussler, a bioinformatician at the University of California, Santa Cruz, knew the odds were stacked against him. For any segment to stand out during evolutionary analysis, he says, its evolutionary pattern would have to be so remarkable that it “wouldn't happen by chance with a million tosses of the dice”.

To increase the odds, Haussler's group wrote computer programs to search vertebrate genomes for non-coding regions that had changed the least over time. The researchers then looked for evidence of accelerated evolution in these regions in the human genome. After a year, Haussler's then postdoc, Katie Pollard, found 49 statistically significant matches (see page 167). They dubbed these 'human accelerated regions', or HARs. “I didn't actually think we would find any,” Haussler says, observing that the likelihood of any region standing out among the tens of thousands tested was very small. “Weird things happen by chance when you do so many tests,” he adds. But, by looking at more species and developing more rigorous tests, he and his colleagues honed the technique. One of the 49 HARs — now known as HAR1 — repeatedly stood out. The group set out to determine its structure and function.

The online genome browser of the Santa Cruz campus was integral to the team's success. It was not only of use for comparing genomic sequences, but provided the first of several lucky breaks. One day, while browsing in the region of the HAR1 sequence, Pollard noticed that a genome-wide bioinformatic scan by another postdoc in the lab, Jakob Pedersen, had predicted a structural RNA gene at the same location. Biochemical analyses established that HAR1 does encode a structural RNA.

The team's luck continued. A visiting Belgian colleague, Pierre Vanderhaegen from the Institute of Interdisciplinary Research in Human and Molecular Biology in Brussels, had samples of human embryonic brain tissue from different stages of development, and agreed to test these for HAR1 expression. The results of his analysis confirmed that HAR1 was expressed when the cerebral cortex was being laid out. “That was the moment we knew HAR1 was involved in brain development,” says Haussler. “These are the sort of results scientists wish for, but seldom get.

Comparing these findings with similar analyses of macaque brain sections, the group found that in the developing cortex the expression pattern of the region has been highly conserved since hominoids diverged from Old World monkeys about 25 million years ago.

Haussler's group is now trying to determine whether HAR1 binds to a protein or interacts with another RNA. Functional studies in mice will follow, as will investigation of other HARs identified by Haussler and his team.