This chapter introduces tools to characterize genomic features, such as introns, exons, coding sequences, and noncoding regions, and illustrates how differences in such features can be used for de novo gene prediction and sequence annotation. Content sensors (e.g., nucleotide, dinucleotide, and trinucleotide frequencies, strand bias, etc.) were used to identify these genomic features, and signal sensors (e.g., Shine–Dalgarno sequence, Kozak consensus, 5′ and 3′ splice sites, etc.) were used to refine the boundaries between these features along the genomic sequences. Different types of mutation and selection generate different sequence signatures of these genomic features. DNA polymerases involved in the replication of the leading and lagging strands differ in their replication fidelity and mutation spectra, generating various strand-specific nucleotide biases that have been used routinely to identify the origin of replication in bacterial genomes. DNA methylation is another major contributor to mutation bias and has a profound effect on genome evolution. Tracing the evolutionary trajectory of these selection and mutation processes requires a phylogenetic perspective. I illustrate a case where wrong conclusions were arrived at when researchers were not equipped with a phylogenetic perspective.