This paper will provide some examples of misleading annotations with regard to …

• Abstract
• Introduction
• Proteins rich in particular amino acids
• Proteins having several biological activities
• Proteins containing several functional domains
• Conclusion
• Acknowledgements
• References

An example is provided by a protein family encoding putative Asp proteases. It comprises about 30 members sharing the IPR009007 domain for peptidase aspartic (http://www.ebi.ac.uk/InterProScan/). Most of them are predicted to be secretory proteins by PSORT and TargetP. Four of them (At1g09750, At5g07030, At3g54400, and At3g61820) were identified in proteomic studies on the A. thaliana primary cell wall (Borderies et al., 2003; G Boudart et al., unpublished results). Most of them are currently annotated as chloroplast nucleoid or nucleoid DNA binding-related proteins in the NCBI and MIPS databases. These proteins are actually homologous to a tobacco chloroplast nucleoid DNA-binding protein that was shown to have a protease activity (Murakami et al., 2000). The annotation of A. thaliana proteins in the NCBI and MIPS databases thus appears to be misleading. By contrast, all these proteins are correctly annotated as Asp proteases in Uniprot as well as in the A. thaliana TIGR and TAIR databases.