RNAclust - an efficient RNA-clustering pipeline

Genome-wide computational and experimental screens for structured non-coding RNA (ncRNA) genes in different taxa resulted in an extensive list of putative structured ncRNAs. We presented a clustering method which detects homologous secondary structure motifs in a set of sequences by pairwise global sequence-structure alignments [1]. The clustering procedure results in a hierarchical cluster-tree which reflects structural similarities. NcRNAs sharing similar secondary structures are found in the same group, divergent ncRNAs are
separated. We provided a method which extracts reasonable subtrees of high-similar structure motifs automatically [2] and developed an user-friendly framework for semi-automatic detection of high-interesting groups.
There are some applications where it is favorable to use local alignments instead of global ones, eg. if the exact start and end positions of RNAs are not known. We are currently developing the possibility to perform local sequence-structure alignments. The clustering pipeline combined with local sequence-structure alignments could detect local structure motifs on long sequences even if there is no full-length secondary structure.
The pipeline is designed for massive parallel processing to allow genome-wide analysis within a reasonable time scale. The structure-based clustering pipeline is a very useful approach for annotating ncRNAs. The possible applications extends even more with the introduction of local alignments.