HMM based ITS2 annotation

The internal transcribed spacer 2 (ITS2) of the nuclear ribosomal repeat unit is one of the most commonly applied phylogenetic markers. It is a fast evolving locus, which makes it appropriate for studies at low taxonomic levels, whereas its secondary structure is well conserved, and tree reconstructions are possible at higher taxonomic levels. However, annotation of start and end positions of the ITS2 differs markedly between studies. This is a severe shortcoming, as prediction of a correct secondary structure by standard ab initio folding programs requires accurate identification of the marker in question. Furthermore, the correct structure is essential for multiple sequence alignments based on individual structural features. The present study describes a new tool for the delimitation and identification of the ITS2. It is based on hidden Markov models (HMMs) and verifies annotations by comparison to a conserved structural motif in the 5.8S/28S rRNA regions. Our method was able to identify and delimit the ITS2 in more than 30 000 entries lacking start and end annotations in GenBank. Furthermore, 45 000 ITS2 sequences with a questionable annotation were re-annotated. Approximately 30 000 entries from the ITS2-DB, that uses a homology-based method for structure prediction, were re-annotated. We show that the method is able to correctly annotate an ITS2 as small as 58 nt from Giardia lamblia and an ITS2 as large as 1160 nt from humans. Thus, our method should be a valuable guide during the first and crucial step in any ITS2-based phylogenetic analysis: the delineation of the correct sequence. Sequences can be submitted to the following website for HMM-based ITS2 delineation:

ITS2 database update III (with Dept. of Bioinformatics)

The internal transcribed spacer 2 (ITS2) is a widely used phylogenetic marker. In the past, it has mainly been used for species level classifications. Nowadays, a wider applicability becomes apparent. Here, the conserved structure of the RNA molecule plays a vital role. We have developed the ITS2 Database ( which holds information about sequence, structure and taxonomic classification of all ITS2 in GenBank. In the new version, we use Hidden Markov models (HMMs) for the identification and delineation of the ITS2 resulting in a major redesign of the annotation pipeline. This allowed the identification of more than 160,000 correct full length and more than 50,000 partial structures. In the web interface, these can now be searched with a modified BLAST considering both sequence and structure, enabling rapid taxon sampling. Novel sequences can be annotated using the HMM based approach and modelled according to multiple template structures. Sequences can be searched for known and newly identified motifs. Together, the database and the web server build an exhaustive resource for ITS2 based phylogenetic analyses.

Reference: Koetschan, C., Förster, F., Keller, A., Schleicher, T., Ruderisch, B., Schwarz, R., Müller, T., Wolf, M., and Schultz, J.örg. (2010) The ITS2 Database III—sequences and structures for phylogeny, Nucleic Acids Research, Oxford Univ Press 38, D275–D279.

ITS2 database: