Genome Annotation and Analysis

Abstract

In the preceding chapter, we gave a brief overview of the methods that are commonly used for identification of protein-coding genes and analysis of protein sequences. Here, we turn to one of the main subjects of this book, namely, how these methods are applied to the task of primary analysis of genomes, which often goes under the name of “genome annotation”. Many researchers still view genome annotation as a notoriously unreliable and inaccurate process. There are excellent reasons for this opinion: genome annotation produces a considerable number of errors and some outright ridiculous “identifications” (see 3.1.3 and further discussion in this chapter). These errors are highly visible, even when the error rate is quite low: because of the large numbers of genes in most genomes, the errors are also rather numerous. Some of the problems and challenges faced by genome annotation are an issue of quantity turning into quality: an analysis that can be easily and reliably done by a qualified researcher for one or ten protein sequences becomes difficult and error-prone for the same scientist and much more so for an automated tool when the task is scaled up to 10,000 sequences. We discuss here the performance of manual, automated, and mixed approaches in genome annotation and ways to avoid some common pitfalls. Mostly, however, we concentrate in this chapter on the so-called context methods of genome analysis, which are the recent excitement in the annotation field. These approaches go beyond individual genes and explicitly take advantage of genome comparison.