Analysis of bacterial biotyping datasets with a mass-based phylonumerics approach.

A mass tree, phylonumerics approach, is implemented for the first time with expressed protein mass data acquired in biotyping applications. It is shown, for two separate and diverse bacterial datasets, that the MassTree algorithm can be used to build phylogenetic trees in a single step that mirror those output by biotyping analysis software in the form of a main spectral profile (MSP) dendrogram or alike. Adapted for these applications to accommodate higher mass inputs and large mass error tolerances for pairwise matching, the mass tree algorithm and approach offers an alternative to commercial biotyping platforms by utilizing datasets acquired from any mass spectrometer without the need for specialized and expensive software.

Mechanisms of antiviral resistance in influenza neuraminidase revealed by a mass spectrometry based phylonumerics approach.

A mass based phylonumerics approach is shown to be able to investigate the origins of the emergence of antiviral resistance mutations in influenza neuraminidase through a global view of mutational trends. Frequent ancestral and descendant mutations that precede and follow the manifestation of antiviral resistance mutations are identified in N2 neuraminidase. The majority occur in the head region around the active site and drive hydrophilicity changes, primarily through the incorporation or loss of hydroxyl groups. These increase or reduce the accessibility of the site to the bulk solvent. The most frequent ancestral mutations that occur on at least two occasions are I/L307M, G/A414S/T, I312T, I/L307S, P386S and S367N; the latter introducing a glycosylation site. The most frequent descendant mutation, after incorporation of an antiviral resistance mutation, is D/E401G/A. Together with others observed, this restore the protein's hydrophobicity about the active site region that limits entry of a sialic acid or inhibitor molecule and reduces viral fitness. The results of this global in silico phylonumerics study demonstrate that evolutionary mechanisms and functionally linked or compensatory mutations, remote in the sequence or structure, can be identified and interrogated at a molecular level.

Mass-Based Protein Phylogenetic Approach to Identify Epistasis.

A mass-based protein phylogeny method, known as phylonumerics, is described to build phylogenetic-like trees using a purpose-built MassTree algorithm. These trees are constructed from sets of numerical mass map data for each protein without the need for gene or protein sequences. Such trees have been shown to be highly congruent with conventional sequence-based trees and provide a reliable means to study the evolutionary history of organisms. Mutations determined from the differences in the mass of peptide pairs across different mass sets are computed by the algorithm and displayed at branch nodes across the tree. By definition, since the trees display a phylogeny representing expressed proteins, all mutations are non-synonymous. The frequency of these mutations and a mutation score based on a sum of these frequencies weighted based upon their position to the root of the tree are output. The algorithm also outputs lists of pairs of mutations separated along interconnected branches of the tree. Those which co-occur or which occur consecutively, or near consecutively, and that are separated by a distance less than the average distance for all mutation pairs, are putatively assigned to be epistatic pairs. These pairs are examined further with a focus on non-conservative substitutions given their importance in driving structural and functional change and protein and organismal evolution. The application of the method is demonstrated for the H3 hemagglutinin protein of type A human H3N2 strains of the influenza virus. The most frequent ancestral mutations within epistatic pairs occur within antigenic site domains while the descendant mutations occur either at other antigenic sites or elsewhere in the protein. Both predominate at reported glycosylation sites. The results for this protein further support a "small steps" evolutionary model for the influenza virus where non-conservative mutations that involve the least structural change are favored over those involving substantive change, which may risk the virus's own extinction.