Predictive Power of Phylogenetics Analysis

Phylogenetics is practiced by most as a historical science; however, several researchers noted that aspects of the tree shape may be used in predicting future genetic lineages of influenza against which it is important to design vaccines [GRE04]. Notable among these assertions are the studies in the shape of influenza A phylogeny as viewed through the hemagglutinin (HA) gene

Fig. 2.7. Two character optimizations on the for hemagglutinin (HA) sequences representing 2,358 isolates of influenza A, with an influenza B outgroup at the root. The top tree has an optimization of the character "HA antigenic subtype". The lower tree depicts optimization of the character "host". Character data was drawn from Genbank (www.ncbi.nlm.nih.gov) and the Influenza Sequence Database (www.flu.lanl.gov). Optimizations and tree graphics were made with Mesquite (www.mesquiteproject.org). For better visualization contact the authors for files in scalable pdf format

Fig. 2.7. Two character optimizations on the for hemagglutinin (HA) sequences representing 2,358 isolates of influenza A, with an influenza B outgroup at the root. The top tree has an optimization of the character "HA antigenic subtype". The lower tree depicts optimization of the character "host". Character data was drawn from Genbank (www.ncbi.nlm.nih.gov) and the Influenza Sequence Database (www.flu.lanl.gov). Optimizations and tree graphics were made with Mesquite (www.mesquiteproject.org). For better visualization contact the authors for files in scalable pdf format

[BUS99, FER02]. The HA gene codes for a surface glycoprotein of the virion responsible for binding to sialic acid on host cell surface receptors. At a ge-nomic level, lineages of influenza are constantly changing due to mutation that occurs at high rates in RNA viruses. Extinction of evolutionary lineages of viruses to which hosts have become immune or when susceptible hosts are in short supply is common [GRE04]. This process of constant replacement of influenza lineages produces a characteristic coniferous shape to a phylogeny reconstructed from HA sequences[BUS99]. The "conifer" metaphor refers to the hypothesis that influenza HA is constantly changing but there is limited diversity at any time [FER02]. Thus an influenza HA tree appears to be formed by addition of strains to the apex of the tree's trunk that contains the contemporary "infectious" viruses rather than more basal presumably "extinct" lineages to which hosts are immune. Other groups of researchers have used the assumption that there is limited influenza A diversity at any one time to downplay the utility of phylogenetic approaches [PLO02]. As an alternative to phylogenetics, which they consider difficult, these groups make predictions based on size of various clusters of related isolates, termed "swarms" [PLO02]. Several groups, whether using trees or swarms, have identified pu-tatively dominant strains of influenza to predict the genetic makeup of future viral populations [PLO02] [BUS99].

If these assumptions were never violated, the diversity of a previous year's flu season could be assessed, forthcoming strains predicted, and thus used to inform vaccine design. In practice, the CDC uses a mixture of viral strains comprised of H1N1 and H3N2 of influenza A and an influenza B virus. For example, the 2005-2006 vaccine was based on A/New Caledonia/20/99 (H1N1), A/California/7/2004 (H3N2), and B/Shanghai/361/2002 viruses [PAL06]. Notably the H5N1 strain (or any of the other avian strains with potential to infect humans) is currently not considered in the vaccine that is seasonally administered to civilians in the United States.

The ability to predict influenza viral strains that will affect human and animal populations is important. However, prediction methods and experimental designs that are relevant to those methods are in their infancy. Current surveillance programs are focused on detection of antigenically novel strains. As such, surveillance programs are not designed as ecological experiments to quantitatively measure strain-specific incidence and cluster size. Furthermore, the current sample of influenza diversity may be biased by partial genomic sequencing, differences in effort within various geographic and political boundaries, focus on certain subtypes of interest, and differential efforts over time due to variable public concern. Recent papers using whole genome data have indicated that the conifer like growth assumption of HA-based phylogenies that has been central to predictive models of H3N2 seasonal influenza [BUS99, FER02] may be violated. Full genome analysis of H3N2 has shown that there are multiple co-circulating lineages; some of which may be overlooked by vaccine designs [HOL05, GHE05]. Similarly, our large scale-analysis of 2,359 HA sequences depicts that many subtypes and lineages within subtypes of influenza are circulating and being exchanged among human and animal populations at any one time Fig. 2.7.

Was this article helpful?

0 0

Post a comment