Even this is likely to be a gross underestimate of the true number, because ENCODE hasn't yet looked at every cell type, or every known transcription factor. "When we extrapolate up, it's more like 18 or 19 per cent," says Ewan Birney of the European Bioinformatics Institute in Cambridge, UK, who is coordinating the data analysis for ENCODE. "We see way more switches than we were expecting, and nearly every part of the genome is close to a switch." But - and it is a big but - these findings do not show whether these switches actually do anything useful. Many of them may have played a role in the past, for instance, but are now "disconnected". The other big surprise is that these regulatory regions are widely dispersed throughout the genome, with many lying in the middle of long stretches between genes that were thought to be barren wastelands. More than 95 per cent of the genome may lie within 10,000 base pairs of a switch. "It means that nearly all of the genome is in play for doing something, or if you change it maybe it would have an effect on something somewhere," Birney says. The way in which these switches work is also turning out to be vastly more complicated than thought. One ENCODE study found that individual switches interact with many genes. What's more, most genes are being influenced by numerous switches at the same time. "Almost every gene we look at is physically touching other pieces of DNA, and it's never just one, it tends to be five, eight, 10 sites, and each site in turn has RNAs on it, proteins on it, histones on it," says team member Job Dekker of the University of Massachusetts Medical School in Worcester. This might help to explain one of biology's biggest puzzles: the mystery of the "missing heritability". We know there's a big genetic component to traits and diseases such as height and diabetes, but the genetic variants found so far typically account for only a tiny percentage of this heritability.