Problems and pitfalls of automatic gene annotation, gene collection, domain prediction, and sequence alignment

Because of the following problems within the automatic gene annotation process it is absolutely necessary to manually check and annotate all genes. Almost every myosin gene prediction and its translation produced by the automatic processes contains errors derived from including intronic sequence and leaving out exons, as well as wrong predictions of start and termination sites. It is also absolutely necessary to reanalyse previously published data, as these also contain many sequencing errors (especially sequences produced in the last century) and wrongly predicted translations. Wrongly predicted genes are the main reason for wrong results in domain predictions, multiple sequence alignments and phylogenetic analyses. In the following sections we show and discuss examples to these problems.