Mining for representative regions of virus genuses via protein sequences clustering

Highly conserved sequences (regions or domains) play an important role in phylogenetic analysis. In this study, a novel approach to extract highly conserved sequences as representative regions for the genuses of viruses via protein sequences clustering is proposed. A representative region of one genus is extracted from one distinctive protein sequence group and appears in all members of that genus; a distinctive feature of a protein sequence group formed by a graph-based clustering method is that the sequences within that group appear in all members of one genus but none of the other genuses. Experimental results showed that there were 64 representative regions belonging to 20 genuses extracted in this study. This study not only creates a methodology to extract highly conserved sequences for phylogenetic analysis across taxa, but also provides the materials for investigating viral taxonomy from a molecular biology point of view, instead of traditional morphology.