A tailored sparse PCA method for finding vaccine targets against Hepatitis C

We present a tailored sparse principal component analysis approach to identify parts of the Hepatitis C virus (HCV) proteome that may be particularly susceptible to immune pressure and thus may help in the design of an effective vaccine. Considering the highly data-limited HCV NS5B protein, the proposed method reveals two reasonably small sets of potentially vulnerable sites which can serve as new vaccine targets. The potential importance of the inferred sets of sites is emphasized through comparison with available clinical and biochemical data. Specifically, both groups of sites have functional significance, associated with viral replication, and one of the groups is found to be strongly targeted by immune systems of individuals who clear HCV naturally, without drugs.