Sequencing a genome is far easier today than finding all the functional sequences within it. Eukaryotic genes are coded in segments, and the segment boundaries are diverse and changeable in response to cellular regulatory signals. Authentic transcription start and stop sites are hard to locate with certainty. DNA sequences that regulate the timing and location of gene expression are difficult to identify, because control sequences are themselves unremarkable and influenced by context. Initial insights into the lexicon of gene structure and control have come from statistical comparisons within genomes to identify nonrandom patterns. As more genomes are sequenced, evolutionary comparison among genomes will yield considerable insights into organization and function of genome sequences. However, the necessary computational methods have not yet been developed.