Clustering rfam 10.1: clans, families, and classes.

Area of Science:

• Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• The Rfam database organizes non-coding RNAs (ncRNAs) into families based on homology and secondary structure.
• A higher-order organization, termed 'clans,' has been proposed to group related RNA families using experimental and computational data.
• Existing clan assignments may not fully capture all structural relationships among RNA families.

Purpose of the Study:

• To investigate an alternative classification of RNA families within the Rfam database using tree edit distance.
• To compare the results of structure-based clustering with the existing Rfam clan organization.
• To explore novel relationships among RNA families beyond the established clan and class levels.

Main Methods:

• Application of tree edit distance to RNA secondary structures for clustering Rfam families.
• Comparative analysis of the generated structural clusters against existing Rfam clans.
• Identification of RNA families that are dispersed or grouped differently by the new clustering method.

Main Results:

• The structure-based clustering partially recovers some existing Rfam clans.
• A significant portion of Rfam clans are not recovered, with families dispersed into larger, biologically relevant clusters.
• These larger clusters align with known RNA classes, such as small nucleolar RNAs (snoRNAs), microRNAs (miRNAs), and CRISPR RNAs (CRISPRs).

Conclusions:

• Structure-based clustering provides a complementary view of RNA family relationships in Rfam.
• This approach can reveal evolutionary and functional connections not captured by the current clan system.
• The findings suggest that RNA secondary structure is a powerful feature for discovering novel ncRNA relationships and classifications.