Its specific role in these processes remains largely unstudied. Here we demonstrate that Irc20 has E3 activity, present genetic results that implicate a role for IRC20 in transcriptional regulation, and demonstrate physical interactions of Irc20 with Cdc48 and SUMO. Biological databases store, organise and share ever-increasing quantities of data. In addition to storing raw biological data, such as protein sequences, many databases aim to attach Ergosterol textual annotation to a given database entry. This textual annotation provides a mechanism to convey understanding of the underlying biology, providing information such as protein function and subcellular location. In describing the current knowledge about the database entry, textual annotations can form the foundations for further research emphasising their crucial role in biological databases. The quality and correctness of textual annotations inevitably varies between databases and entries. This can depend on many factors, such as: the current evidence supporting the function of the protein; the Chloroquine Phosphate curation and review process; and the curators’judgement in extracting information from biomedical literature. The kind of metadata describing annotations also varies between databases and entries, limiting the ability to compare them. For example, the source and last updated date of a Gene Ontology annotation is not always apparent. At the highest level, we can distinguish between two types of annotation curation process: manual curation and automated curation. It is generally held that manual curation is of higher quality and correctness than its automated counterpart. This is mainly because expert curators have the ability to access, evaluate and interpret a wide range of scientific literature as a source of information for annotations. However, automated annotation pipelines, such as UniRule, provide greater annotation coverage and more regular updates, as annotations are often transferred from existing annotations. Database sizes are continuing to expand at an exponential rate, resulting in a continued and growing reliance on automated curation. Identification of textual annotation that could be of interest in the curation process is often based upon biological sequence; sequences that share properties, such as sequence similarity, are more likely to share a similar function and attributes. Given a strong sequence similarity, it is reasonable that annotations may be copied verbatim between entries, i.e. sentences are subjected to reuse. Therefore, annotations are often based purely, or in part, on existing annotations. It is also becoming an increasingly common practice for manual curators to use existing annotations within their curation process; either from annotations within the existing database or from external databases. If a database lacks formal provenance and metadata, it may mean that it is not possible to identify the original source of an annotation. Given this, the extracted textual annotation may have also previously been copied from other entries. Should the original source of a textual annotation be found to be erroneous, there is no clear way of identifying where it has propagated to. A number of studies have explored textual annotation quality, however, very limited work has explicitly explored textual annotation propagation and its link to correctness. One such study explores the usage of association rules to detect possible erroneous annotations. This study, performed on the Swiss-Prot database, focused primarily on the annotation within the feature table; free text annotation were 
omitted from the analysis. The reason for this omission was given as ” is not easily machine-parseable”.