Logical foundations of W-Curves used for phylogenetic analysis: An investigation of W-Metrics.

摘要

We have defined a new metric (the W-Metric) on W-Curves for the use in the development of phylogenetic trees based on the argument that similarity between the W-Curves (that is, similarity between the underlying DNA strands) is the most logical basis for development of phylogenetic trees that reflect genetic kinship. The W-Curve was developed here at Illinois Institute of Technology as a tool for efficiently visualizing long genomic sequences.; We have tested our metric on published data about genome sequences for mammals where the phylogenetic trees are known and also on genetic data from the HIV virus. The trees resulting from the mammals data set strongly support the adequacy of this method to map out the classification of species. We have tested the efficiency and effectiveness of the W-Metric in building the trees for mammalian gene data, HIV-1 sequence data and the Bacitracin Synthetase sequence data. In the analysis, the results have been compared and contrasted with the ones obtained from other approaches. Subsequent general evaluation of procedures for doing phylogenetic analysis using W-curves lead to identify some areas in which improvements can be made.; The W-Curve method is computationally attractive since it provides a fast way of visually detecting special or repeating patterns, approximating the similarity. The W-Metric allows us to do in depth analysis by using W-Curves to support the construction of phylogenetic trees. However, the auto-regressive nature of the W-Curve does not lend itself to accurately quantify dissimilarity using extensions of existing means like percent difference or Euclidean metric approximation.; This research uncovered a problem with estimating dissimilarity between two W-Curves which is inherent to the concept of the auto-regressive function itself, namely the concern about overestimation of branch length. One way to investigate this problem more closely would be to attempt to produce the W-Curve of two or three medium or short sequences which differ just by one base. The dissimilarity has been defined by many phylogeny programs as the percent difference. The iterative function would not allow an almost zero dissimilarity between two W-Curves of which representative DNA sequences differ by only one base.; Gap Stripping resolution using a consensus algorithm, as we implemented in this research, is also of major importance since it provides us some insurance that biologically meaningful information is not discarded before it has served its purpose.