The Lost Denominator

The Washington Post had an interesting article by Rick Weiss on the unraveling of the Chimpanzee genome which had a little problem with the math that it shares with many popular science articles. The conclusion of the article is that the chimpanzee DNA differs from the human in only 1% of the active DNA.

All told, the two sequences differ by 4 percent. But three-quarters of the differences seem to be in non-functional parts of the genome, suggesting that a mere 1 percent variation makes all the difference.

It's pretty clear that if you take 1/4 of 4% you get 1%. But from what the article says we can't say if that's the right approach. Suppose there are 3 billion letters in the genome. Then the article is saying that there are about 120 million differences in the total genome. About 90 million of these are in the non-functional part of the genome and 30 million are in the functional part. So far we're on pretty solid ground. But can we calculate the fraction of differences in the functional part?

What we don't know is what the size of the functional part is. I'm not a specialist in the field, but this article suggests that in one particular area about 39% of the genes are non-functional (and not all of the DNA sequence is in genes).

It may be that there is some new and different definition of non-functional that's being used in this latest article but I've certainly seen other articles indicating that most of our DNA is non-functional. It would sure have been nice if the author had thought to describe the change in the size of the denominator when he implied that the ratio of changes in the functional area of the genome was only 1%. I'd be a lot more willing to believe it.