hapflk

We are happy to introduce a new version of hapflk that improves how the population tree is estimated and consequently the estimation of the population kinship matrix.

The improvement concerns mainly situations where no outgroup information is available. In the previous hapflk versions, in such cases we used a midpoint rooting approach, which could lead to biased estimations of the populations kinship matrix, as exemplified in the recent paper by Gautier (2015). We now implement a maximum likelihood-based approach to (i) identify the branch in the unrooted tree where the root is located and (ii) optimize the root placement within that branch. The rationale is to adjust branch lengths to observed population heterozygosities.

The new implementation makes it possible to use hapflk in datasets with only two populations and no outgroup information, which was not possible with previous releases. It also corrects a bug with the --keep-outgroup option where the root was previously ill-placed, in particular in datasets with few populations.

All of this has been reimplemented in python, so that performing these procedures with hapflk no longer requires R nor its packages ape and phangorn.

• Gautier M. (2015) Genome-Wide Scan for Adaptive Divergence and Association with Population-Specific Covariates. Genetics. 10.1534/genetics.115.181453