Phylogenetic inference from homologous sequence data: minimum topological assumption, strict mutational compatibility consensus tree as the ultimate solution.

Background For the purposes of phylogenetic inference from molecular data sets many different methods are currently offered as alternatives for researchers in phylogenetic systematics. The vast majority of these methods are based on specific topological assumptions relating to the resultant genealogical tree. Each of these has been shown to perform effectively in special conditions and for specific data sets while yielding less reliable results in other instances. Moreover, the majority of the methods include information from homoplastic characters in spite of a universally accepted agreement in their ineffectiveness for phylogenetic inference, which may often lead to inaccuracy and inconsistency. As an alternative to such methods, a strict mutational compatibility consensus tree building method as a universally applicable and reliable method is reported.Results The analysis of a data set from a previously published experimental phylogeny demonstrates the accuracy of the strict mutational compatibility consensus tree building method and illustrates its potential for obtaining unambiguous and precise results with full resolution.Conclusion The universal applicability of a simplified compatibility method in its algorithmic form for phylogenetic inference is described. Firstly, dismissal of topological assumptions creates a general potential for agreement of inferred with true phylogeny. Second, exclusion of irregular characters from analysis repeatably enables construction of consistent phylogeny. Third, a direct calculation of bootstrap proportion values for individual nodes of the resulting tree is possible rather than their empirical estimation. Finally, guidance is given for empirical assessment of the sample size necessary for full genealogical resolution and significant bootstrap proportions.Reviewers This article was reviewed by Yuri I. Wolf (nominated by Eugene Koonin), Arcady Mushegian and Martijn Huynen.