When biotech firm Celera was attacked for its cockiness in vowing to sequence the human genome before university and government scientists did, the company had a powerful comeback: hey, we sequenced the fruit fly first, remember? Last year Celera and UC, Berkeley, finished sequencing the chemical "letters" (A's, T's, C's and G's) in the DNA of Drosophila. But when mathematician Samuel Karlin of Stanford University compared Celera's sequences with those worked out and corroborated in experiments, he found "significant discrepancies," he reports in Nature: 45 percent of the fly genes contained serious errors (like letters in the wrong place). Celera admits its sequence "is still a work in progress." And the human genome, which the public project and Celera were hellbent on finishing by this spring? It, too, may contain substantial errors. Oops.