Genome annotation

Genome annotation assigns features and evidence-backed interpretations to a genomic sequence. It can locate genes, repeated elements and regulatory regions, while associating predicted products with possible functions. Structural annotation asks where features lie; functional annotation asks what they do. Both depend on evidence and can change when new data arrive. An apparently missing gene can reflect incomplete assembly or annotation rather than a genuine evolutionary loss. ^[ref]

Gene annotation identifies the structure and attributes of a gene in a reference sequence. It records features such as exons, splice forms and coding regions, often combining RNA evidence with sequence comparison. A supported transcript structure does not automatically establish a known biochemical function for its product. Distinguishing structural evidence from functional interpretation prevents a prediction from being mistaken for a fully characterized gene. ^[ref]

Gene prediction infers candidate genes and their boundaries from genomic sequence and supporting data. Models can recognize coding patterns and splice signals, while transcript and comparative evidence help constrain possible structures. Predictions can miss short genes, confuse duplicated regions or propose unsupported splice forms. Experimental transcript evidence can improve a gene model, but observation of a transcript does not alone establish that it produces a functional protein. ^[ref]

A gene model is a structured representation of a gene, including its genomic location and proposed transcript structures. One model can include several splice isoforms with different exon combinations. Changing the reference sequence or adding RNA sequencing evidence can revise boundaries and coding predictions. The model represents current evidence and inference rather than a claim that every included transcript has been observed in every tissue. ^[ref]