0.3.1
- requires gtdblib >= 1.12.1
- messages from gtdblib use the GenomeTreeTk log format, are written to genometreetk.log, and are suppressed by --silent
- errors reported by gtdblib (e.g. a taxon without a rank prefix) are logged with exit code 1 instead of a traceback

0.3.0
- replaced biolib with gtdblib (>= 1.12.0); packaging now uses setuptools and requires Python >= 3.7
- bootstrap: MSA replicates are created from a shared numpy matrix (about 7x faster and 4x less memory for 200,000 x 5,000 MSAs)
- bootstrap: added --seed; replicates are identical for any number of CPUs
- bootstrap: reruns safely resume interrupted runs; FastTree failures are reported instead of ignored
- bootstrap: fixed --boot_dir with MSA replicates; --msa_only no longer requires FastTree
- bootstrap and jk_markers: support values are calculated with gtdblib, which keeps the rooting and taxon labels of the input tree and writes whole-number support
- jk_markers: FastTree failures are reported instead of ignored
- combine: handles nodes without support values (e.g. the root) and decimal support values
- errors now exit with a non-zero code
- fixed pull writing invalid taxonomy strings for leaves without labelled ancestors
- fixed derep_tree (required --msa_file and failed on Python 3) and arb_records (failed on Python 3)
- removed unused modules (infer_workflow, markers, ncbi, aai)
- added end-to-end tests of the commands in tests/

0.2.3
- added sh_uf_hack command

0.2.2
- fixed bug with filtered MSA columns being written out in a random order within each marker
- now generating final_mask.txt which is the mask after filtering and subsampling

0.2.1
- modified bootstrap method to support use of FastTreeMP with multiple CPUs

0.2.0
- added msa_filter method

0.1.9
- added support to create MSA replicates without inferring replicate trees
- added support to infer trees from MSA replicates if they do not already exist

0.1.8
- better support for doing bootstrap trees across multiple servers

0.1.7
- updated propagate command to work with new Gxyz style genome IDs
- updated 'bootstrap' command to not require input tree

0.1.6
- fixed more merging issues

0.1.5
- fixed mering issue

0.1.4
- added prune method

0.1.3
- modified main.py to have Python 3 print statements

0.1.2
- updated shebang to Python 3

0.1.1
- added CM data files to repository

0.1.0
- initial migration to Python 3

0.0.55
- added rm_support
- bootstrap command will not look for files with the extension '.treefile'

0.0.54
- modified PD calculations when ingroup does not encompass root node; this
  makes PD a rooted tree calculation, but this is probably best for most
  situations
- simplified interface to PD calculations to remove notion of representative genomes

0.0.53
- better handling of support values when 'stripping' a tree

0.0.52
- modified cluster_de_novo to ensure all genomes as assigned to the closest type genome
- fixed synonym table: species were being indicated as synonyms of multiple species (e.g. Shigella boydii)

0.0.51
- added --rnd_type_genome option for evaluating impact of type genome selection on de novo clusters

0.0.50
- added --gamma option to bootstrap command so it isn't automatically used

0.0.49
- used for GTDB r89
- added QC exception list to qc_genomes command
- added species classification exception list to qc_genomes command
- added species classification exception list to select_type_genomes command
- added species classification exception list to cluster_named_types command
- added GTDB type genomes exception list to select_type_genomes command
- added species type genome radius information to file indicating species clusters
- simplified interface to cluster_user as species ANI radius information is now in cluster file
- stripped 'Candidatus' from all species names

0.0.48
- added type species of genus information for resolving priority of species with same publication date
- changed {manual_curation} to {unresolved}
- changed determination of genus name for de novo species clusters to consider most common genus name in cluster
- added checks to ensure genomes marked as assembled from type material or NCBI representative genomes as not actually a subspecies

0.0.47
- initial implementation of filtering of percentage of bac120 or ar122 marker genes

0.0.46
- initial type genome workflow for GTDB r89

0.0.45
- improved dereplication workflow for GTDB r89

0.0.44
- add method to dump 5S, 16S, and 23S sequences to file

0.0.43
- improvements to code for assigning genomes to canonical genome set 

0.0.42
- initial implementation of code to assign genomes to canonical genome set 

0.0.41
- fixed bug with parsing contig length from SSU file

0.0.40
- fixed issue with rerooting trees with bootstraps between [0, 1]

0.0.39
- added GTR model when inferring SSU or LSU trees

0.0.38
- new method for selecting GTDB representatives

0.0.37
