Edit on GitHub

Introduction

Whale provides tools for (genome-wide) amalgamated likelihood estimation (ALE) under a DLWGD model, which is an approach to infer reconciled gene trees and parameters of a model of gene family evolution given a known species tree. The ALE approach takes into account uncertainty in the gene tree topology by marginalizing over all tree topologies that can be amalgamated from the so-called conditional clade distribution (CCD). This CCD can be constructed from a sample of the posterior distribution of tree topologies (which can be obtained using any standard software for Bayesian phylogenetics, see below).

More specifically, this library can be used to

  • Statistically assess hypothetical whole-genome duplication (WGD) events in a species phylogeny.
  • Infer lineage-specific gene duplication and loss rates for a species phylogeny.
  • Infer high-quality (reconciled) gene trees given a known species tree using Bayesian gene tree reconciliation cf. Szöllősi et al.
  • Conduct Bayesian orthology inference.
  • All of the above at once.
Note

This library implements the DL and DLWGD models. It does not implement models of gene family evolution that take into account horizontal gene transfer, incomplete lineage sorting or gene conversion.

Installation

You will need julia-1.x. Fire up a julia REPL by typing julia at the command line enter the package manager interface by typing ], and execute add https://github.com/arzwa/Whale.jl.

Data preparation

To perform analyses with Whale, you will need

  1. An ultrametric species tree, with ideally branch lengths in geological time (since this allows straightforward interpretation of parameter estimates.)
  2. A bunch of ALE files, which summarize the conditional clade distributions (CCDs) for the same bunch of gene families. These can be obtained from a sample of the posterior distribution of gene trees (or bootstrap replicates, if you prefer) using the ALEobserve tool.
Note

Gene IDs should be prefixed by the name of the species to which the gene belongs as used in the species tree. For example if Arabidopsis thaliana is represented by ATHA in the species tree newick file, then the genes should be prefixed with ATHA_, e.g. ATHA_AT1G05000.

Note

If one wants to conduct gene tree reconciliation using Whale assuming known gene trees as input, one can simply construct a trivial CCD for each gene tree using ALEobserve (i.e. construct a CCD for a 'sample' consisting of a single gene tree).

Note

Analyzing CCDs (ALE files) with a very large number of clades or for very large families can be prohibitive computationally. It is therefore generally advisable that large orthogroups are filtered out based on some criterion (for example using the script orthofilter.py in the scripts directory of the Whale repository). To filter out families with very large numbers of clades in the CCD (which reflects that there is a lot of uncertainty in the gene tree), the scripts ccddata.py and ccdfilter.py can be used. This is a rather ad hoc filtering procedure, but can be useful to filter out families that trouble the analysis.

Warning

Most analyses in Whale assume that for each family, there is at least one gene in both clades stemming from the root of the species tree. The likelihood in Whale is the conditional likelihood under this assumption. This is to rule out the possibility of de novo gain of a gene family along a branch of the species tree. The orthogroup data should therefore always be filtered to be in accordance with this criterion. This can also be done using the orthofilter.py script.

References

Whale.jl is developed by Arthur Zwaenepoel at the VIB-UGent center for plant systems biology (bioinformatics & evolutionary genomics group). If you use Whale, please cite:

Zwaenepoel, A. and Van de Peer, Y., 2019. Inference of Ancient Whole-Genome Duplications and the Evolution of Gene Duplication and Loss Rates. Molecular biology and evolution, 36(7), pp.1384-1404.

The methods in Whale are heavily inspired by previous work done by other researchers. If you use Whale, consider citing the following two particularly important studies:

[ALE] Szöllősi, G.J., Rosikiewicz, W., Boussau, B., Tannier, E. and Daubin, V., 2013. Efficient exploration of the space of reconciled gene trees. Systematic biology, 62(6), pp.901-912.

[DL+WGD model] Rabier, C.E., Ta, T. and Ané, C., 2013. Detecting and locating whole genome duplications on a phylogeny: a probabilistic approach. Molecular biology and evolution, 31(3), pp.750-762.