Gene-Finding Approaches for Eukaryotes

The goal of this paper is to introduce the methods commonly used for predicting protein-coding regions in eukaryotic DNA, primarily for the benefit of those not familiar with the topic. This is not meant as a comprehensive review, nor do I describe in detail the underlying mathematical formalism. Those seeking additional information are encouraged to read some recent reviews (Fickett 1996; Claverie 1997; Burge and Karlin 1998; Haussler 1998). Most of the papers in this issue from the recent Genome Annotation Assessment Project (GASP) attempt to identify protein-coding genes using one or more of the methods I describe. I do not assess the success of the different methods, as that is done in the accompanying paper (Reese et al. 2000) and by each paper individually.

There are two important aspects to any program for gene identification: one is the type of information used by the program, and the other is the algorithm that is employed to combine that information into a coherent prediction. Three types of information are used in predicting gene structures: “signals” in the sequence, such as splice sites; “content” statistics, such as codon bias; and similarity to known genes. The first two types have been used since the early days of gene prediction, whereas similarity information has been used routinely only in recent years. One of the reasons that the accuracy of gene-prediction programs have improved in the last few years is the enormous increase in the number of examples of known coding sequences. This much larger sample size allows for more reliable statistical measures to be developed, as well as a much greater likelihood of encountering a gene that is related to one that has been identified previously.