<P>It’s just a new genescan computer algorithm being applied. Bioinformatacists are constantly tweaking the computer programs that scan the genome sequence and try to predict all the protein-encoding genes. At the release of the draft sequence, the estimate was ~30,000. Then it went up to 40,000, then 50,000, then back to 30,000, now down to 20,000. There are some people that seem to think that this new estimate of 20,000 genes is more accurate because the genome sequence is now quite “polished” (as we geneticists say). But I wouldn’t bet on it. We are not going to know the true number until we have hands-on experimentally verified them all, and that’s going to take a long looooooong time.<P>