Peptide poset mining for antiviral target regions in Human Rhinovirus

Human Rhinovirus(HRV) causes benign upper respiratory disease and exacerbation of infection like asthma, often leading to more severe chronic pulmonary illness specially in young children and immune-suppressed people. Precise identification of antiviral target regions is a necessity for effective drug design, but that falls through by biological methods for high level of diversity of the region. Moreover, in vitro bioculture processes are bounded by fixed time quanta. We observe that intervals of successive occurrences of each amino acid in a peptide sequence form a poset. Entropies of posets of individual amino acids and of their combinations represent sequence feature. We propose a computational method, Peptide Poset Mining(PPM), for specific feature discovery in a polypeptide sequence. Experiments on several serotype sequences of HRV reveals that PPM is capable to isolate the replication(P2) and assembly(P3) regions of the pathogen. The region VP4 has high potential for next generation pan-serotype HRV vaccine for its more conserveness over VP1, but it is often untraceable by computational methods due to its small length. Our method precisely identifies VP4.