Statistical Inference of Phylogenetic Networks

The Liu and Bonito labs have created a new method for scalable statistical inference of phylogenetic networks from large-scale genomic sequence datasets. The new method has been published as part of peer-reviewed proceedings of RECOMB-CG 2018. The publication can be accessed here.

In related work, members of the Liu lab have developed a novel non-parametric/semi-parametric method for statistical support estimation. The approach has initially been applied to classical problems in computational biology and bioinformatics. The work was also published as part of peer-reviewed proceedings of RECOMB-CG 2018. The publication can be accessed here. An extended journal paper has been submitted to Algorithms for Molecular Biology.

The two-reticulation species phylogeny was inferred by FastNet(MPL) on a yeast genomic sequence dataset from Salichos and Rokas’s 2013 study. Branch lengths (blue text) are given in coalescent units. Reticulation edges (dashed red lines) are annotated with admixture frequencies (red text). The two-reticulation species network was preferred to phylogenetic hypotheses involving fewer reticulations. For reference, the putative whole genome duplication event described by Gabaldon et al. (2013) would be placed on the branch incident to the MRCA of the sampled Saccharomyces taxa and Candida glabrata (i.e., the branch with length 0.22). The Dendroscope software package was used in the process of producing the illustration.

01.22.19