Annotation of the Drosophila genome

The publication of the genome of the fruitfly Drosophila by Celera Genomics last year generated considerable excitement1. The initial release predicted 14,226 proteins, including about 550 that arise as the result of alternative splicing1. Here we compare the annotation of the Celera Drosophila genome with known protein sequences lodged in a databank and find that although there is substantial agreement for many gene sequences, there are also numerous and significant discrepancies. These uncertainties call for caution in interpreting predicted Drosophila genes.