EPA-ng version 0.3.0

WARNING This version has a fatal error in the MPI version, please use v0.3.2 instead!

EPA-ng is now peer reviewed! Please cite the following paper when using it in your work:

Pierre Barbera, Alexey M Kozlov, Lucas Czech, Benoit Morel, Diego Darriba, Tomáš Flouri, Alexandros Stamatakis;
EPA-ng: Massively Parallel Evolutionary Placement of Genetic Sequences,
Systematic Biology, syy054, https://doi.org/10.1093/sysbio/syy054

This version introduces root preservation of the reference tree, compulsory specification of the model string/parameters, as well as some minor bug fixes and improvements. This is also the first verion I don't consider beta anymore, as the program has been very stable over the last months across multiple projects and datasets.

Interface changes

• --model parameter is now compulsory and works identically with the same option in raxml-ng
• added --precision argument to specify output floating point precision

Feature changes

• added a check to disallow multifurcations in the reference tree
• output floating point precision is now of uniform length (can be set via --precision)
• EPA-ng now preserves rooting, meaning that if you specify a rooted tree as the reference, that same tree will be the tree in the output. Unrooted trees are still used internally, but the placement results are translated to the rooted tree as they are written to file

Bugfixes

• added code to catch sequences that are all gap and fail when found
• fixed possible failure when empty work structures would have been merged
• fixed a bug where pll scalers were copied incorrectly
• fixed a bug sometimes leading to incorrect placement distal lengths

Misc changes

• switched to cmake build for pll/pll-modules
• improved some error messages
• added test data and link to test data used in the paper
• updated citation info