Multiple sequence alignment accuracy and phylogenetic inference (original) (raw)

Multiple Sequence Alignment in Phylogenetic Analysis

Molecular Phylogenetics and Evolution, 2000

Multiple sequence alignment is discussed in light of homology assessments in phylogenetic research. Pairwise and multiple alignment methods are reviewed as exact and heuristic procedures. Since the object of alignment is to create the most efficient statement of initial homology, methods that minimize nonhomology are to be favored. Therefore, among all possible alignments, the one that satisfies the phylogenetic optimality criterion the best should be considered the best alignment. Since all homology statements are subject to testing and explanation this way, consistency of optimality criteria is desirable. This consistency is based on the treatment of alignment gaps as character information and the consistent use of a cost function (e.g., insertion-deletion, transversion, and transition) through analysis from alignment to phylogeny reconstruction. Cost functions are not subject to testing via inspection; hence the assumptions they make should be examined by varying the assumed values in a sensitivity analysis context to test for the robustness of results. Agreement among data may be used to choose an optimal solution set from all of those examined through parameter variation. This idea of consistency between assumption and analysis through alignment and cladogram reconstruction is not limited to parsimony analysis and could and should be applied to other forms of analysis such as maximum likelihood.

REVIEW Multiple Sequence Alignment in Phylogenetic Analysis

The impact of sequence parameter values on phylogenetic accuracy

An accurately inferred phylogeny is important to the study of evolution. Factors affecting the accuracy of an inferred tree can be traced to several sequential steps leading to the inference of the phylogeny. We have examined here the impact of some features of nucleotide sequences in alignments on phylogenetic (topological) accuracy rather than any source of error during the process of sequence alignment or choice of the method of inference (as is usually done). Specifically, we have studied (using computer simulation) the implications of changing the values of the following five parameters, individually and in combination: sequence length (l), nucleotide substitution rate (r), nucleotide base composition (θ), the transition-transversion rate ratio (κ), and the substitution rate heterogeneity among the sites (α). An interesting, and unexpected, result was that κ has a strong positive relationship with phylogenetic accuracy, especially at high substitution rates. This simulation-based work has implications for empirical researchers in the field and should enable them to choose from among the multiple genes typically available today for a more accurate inference of the phylogeny being studied.

The Effect of the Guide Tree on Multiple Sequence Alignments and Subsequent Phylogenetic Analysis

2008

Many multiple sequence alignment methods (MSAs) use guide trees in conjunction with a progressive alignment technique to generate a multiple sequence alignment but use differing techniques to produce the guide tree and to perform the progressive alignment. In this paper we explore the consequences of changing the guide tree used for the alignment routine. We evaluate four leading MSA methods (ProbCons, MAFFT, Muscle, and ClustalW) as well as a new MSA method (FTA, for "Fixed Tree Alignment") which we have developed, on a wide range of simulated datasets. Although improvements in alignment accuracy can be obtained by providing better guide trees, in general there is little effect on the "accuracy" (measured using the SP-score) of the alignment by improving the guide tree. However, RAxML-based phylogenetic analyses of alignments based upon better guide trees tend to be much more accurate. This impact is particularly significant for ProbCons, one of the best MSA methods currently available, and our method, FTA. Finally, for very good guide trees, phylogenies based upon FTA alignments are more accurate than phylogenies based upon ProbCons alignments, suggesting that further improvements in phylogenetic accuracy may be obtained through algorithms of this type.

The Effect of the Guide Tree on Multiple Sequence Alignments and Subsequent Phylogenetic Analyses

Biocomputing 2008, 2007

A framework for phylogenetic sequence alignment

Plant Systematics and Evolution, 2009

A phylogenetic alignment differs from other forms of multiple sequence alignment because it must align homologous features. Therefore, the goal of the alignment procedure should be to identify the events associated with the homologies, so that the aligned sequences accurately reflect those events. That is, an alignment is a set of hypotheses about historical events rather than about residues, and any alignment algorithm must be designed to identify and align such events. Some events (e.g., substitution) involve single residues, and our current algorithms can successfully align those events when sequence similarity is great enough. However, the other common events (such as duplication, translocation, deletion, insertion and inversion) can create complex sequence patterns that defeat such algorithms. There is therefore currently no computerized algorithm that can successfully align molecular sequences for phylogenetic analysis, except under restricted circumstances. Manual re-alignment of a preliminary alignment is thus the only feasible contemporary methodology, although it should be possible to automate such a procedure.

Application and accuracy of molecular phylogenies

Science, 1994

Molecular investigations of evolutionary history are being used to study subjects as diverse as the epidemiology of acquired immune deficiency syndrome and the origin of life. These studies depend on accurate estimates of phylogeny. The performance of methods of phylogenetic analysis can be assessed by numerical simulation studies and by the experimental evolution of organisms in controlled laboratory situations. Both kinds of assessment indicate that existing methods are effective at estimating phylogenies over awide range of evolutionary conditions, especially if information about substitution bias is used to provide differential weightings for character transformations.

Multiple sequence alignment accuracy and phylogenetic inference (original) (raw)

Related papers