The new genes of rice: a closer look

Reports accompanying draft or finished sequences of rice chromosomes and full-length cDNA libraries indicate that between a third and half of the (largely predicted) protein-coding genes of rice might have no identifiable homologs in Arabidopsis and/or other species. The set of apparent 'no-homolog' sequences are predicted to exhibit striking compositional and structural deviations from experimentally verified protein-coding sequences. Here we discuss evolutionary and other implications of the proposed gene set and argue that a more likely answer to the riddle is that a large proportion of the anomalous rice sequences are never translated into functional proteins in vivo, that is, they represent incorrect gene predictions.