v2.3.1

Fixed

• The error occurs when the total number of threads assigned exceeds 4 times the number of jobs during hmmsearch step in the align
  module.

• Unexpected breaks caused by directory being incorporated as inputs in align module.

• The error arose due to behavior changes introduced in pyfaidx 0.9.0.

v2.3.0

Added

• The argument "seqtype" to SeqFileWrapperABC and its subclasses, which allow users to manually assign the seqtype of the inputs
  if the autodetection failed.

• The argument "seed" to tree module providing control to random generator during tree building.

Removed

• Tree displaying at the end of the tree module.

Changed

• Swap the key-value in DataListABC checksums which allow to look up using input checksums instead of input names.

• Use hierarchical loggers to better control the logging behavior. Now core modules (align, filter, tree) will output logs to a
  file [output_dir]/log.txt.

Fixed

• The order of the checksums in HMMMarkerSet generated from different platforms (x86 Linux and M series macos) are different.

• Error when estimating the threads during the align module with only 1 job.

• Error message traceback issue when monitoring the progress with progress daemon.

v2.2.0

Changed

• Use JC instead of GTR for faster phylogenetic inference during filtering step.

• Swap back to FastTree since VeryFastTree stuck occasionally. #35

• Enable --noml option for FastTree during filtering step by default. Users can use --ml in filter module to enable it.

• The hmmsearch step now report only single hit instead of the top hit of each markers. Therefore the final results will be
  single-copy orthologs.

• Use the fractional process indicator in search, filter and tree modules to better monitor the run.

• Move the tree construction comments from the top of the newick file to log.txt to avoid newick file loading failure in some of
  the software.

Fixed

• Fix guess_seqtype failure when having ambiguous codes in input sequences.

v2.1.1

Changed

• Set HMM.name to the filename of the hmm profile to adapt the change on the v12 dataset.

Fixed

• The treeness/RCV computation error when the tree's total branch length is zero.

v2.1.0

Added

• Environment variable PHYLING_DB that can be set to retrieve the database across multiple paths.

• Site concordance factor along with bootstrap value to better interpret branch support.

Removed

• The threads option in SearchHitsManager load method - it run faster in single process/thread.

• Custom bootstrap value assignment in main menu. Now bootstrap was done by UFBoot with 1000 replicates, standardized across all tools.

Changed

• Make SampleList hashable and use it directly in SearchHits instead of just name.

• Remove the selected_MSAs folder and directly link the selected MSAs to the main output folder produced by the filter module.

• Use VeryFastTree to replace FastTree.

• In consensus mode, use GTR and LG models for DNA and peptide data. In concatenation mode, use ModelFinder to find the best model(s)

• Use UFBoot for bootstrapping after tree building by either tool.

Fixed

• The unload data when hmmsearch with multithreads.

• The FileNotFoundError when rerun in a different working directory.

• The missing newline of the treeness file output from filter module.

• Simplify the log when verbose is disabled.

v2.0.0

Added

• phyling download list now will also print out the markersets that have already been downloaded.

• The filter module that calculate the treeness/RCV scores through PhyKIT to filter the uninformative markers. Use
  -n/--top_n_toverr to specify the number of markers you which to use in the final tree building.

• RAxML-NG and IQTree are now available for final tree building.

Removed

• Output option in download module. Now all the BUSCO datasets will be saved in the config folder ~/.phyling/HMM.

• Remove --from_checkpoint feature. Output to the previous output folder will trigger the check and automatically determine the
  rerun status.

• Remove tree building methods UPGMA and Neighbor Joining.

Changed

• Change the align module -m/--markerset behavior. It firstly searches against the given path and the config folder
  ~/.phyling/HMM if the path doesn't exist. Users can also directly specify the markerset name that has already been downloaded
  and saved in the config folder.

• Use timestamp for default output folder in align and tree module.

• Use FastTree to replace the UPGMA for the default tree building method.

Fixed

• Fix the multiprocessing issue.

v2.0.0-beta

Changed

• Use ClipKIT to replace the self-defined function for trimming off the sites that display poor phylogenetic signal.
• Move the MSA concatenate function from align module to tree module. Users who want to try different tree building strategy won't have to rerun the align module again.
• Replace the VeryFastTree with FastTree for stability.

Added

• Check for duplicated sample names.
• Report problematic cds sequences.
• Use checkpoint file to save the hmmsearch results to prevent rerunning the search process when adding/removing samples.

Fixed

• Fix the bug caused by translation from cds sequences with invalid length.
• Fix the bug caused by inconsistent MSA output extension.
• Fix the bug that the trimming function always return peptide MSA if the sequence has no site being trimmed.
• Fix the Python logger issue.

v0.9.0

Fix workflows

v1.1beta Intermediate release

Archive of current code for citation and release, but there are revisions still in progress. Included a CITATION.cff file and some of the framework for BUSCO marker set use direct from unannotated genomes.