Repository navigation
What's Changed
- symlink hsp result if there is only one by @bernt-matthias in #410
- Updates for v2.6.4 by @R-Wright-1 in #421
New Contributors
- @bernt-matthias made their first contribution in #410
Full Changelog: v2.6.3...v2.6.4
- Updated so that split pipeline works with QIIME2 and all current versions of packages
- Added chunk_size for reducing memory usage with epa-ng
- Finally updated the install tests so that they actually test the new split pipeline
- Fixing the EC numbers mentioned in issues 413 and 420:
- Comes from a version mismatch between the predictions made by eggnog and the metacyc reaction mapping taken from HUMAnN
- The fix that I'm adding is not really a full fix. I'm combining the reactions from the old default_files/pathway_mapfiles/metacyc_rxn_to_level4ec.tsv file, the 2.6.0 default_files/pathway_mapfiles/metacyc_rxn_to_level4ec_new.tsv (now renamed to default_files/pathway_mapfiles/metacyc_rxn_to_level4ec_picrust2_2.6.0.tsv) and the HUMANN 3.0.0 metacyc_reactions_level4ec_only.uniref.bz2 (from https://raw.githubusercontent.com/biobakery/humann/3.0.0/humann/data/pathways/metacyc_reactions_level4ec_only.uniref.bz2). This fixes 211 EC numbers that were missing from 2.6.0, but there are still 63 missing (default_files/pathway_mapfiles/missing_ec_from_reaction.txt)
- This replaces default_files/pathway_mapfiles/metacyc_rxn_to_level4ec_new.tsv with default_files/pathway_mapfiles/metacyc_rxn_to_level4ec_v2.6.4.tsv
- At some point in the future I will update the full database and aim to fix this fully by recreating the mapping files with whatever the current MetaCyc version is at the time.
- Removed ete3 as it is no longer maintained. Replaced with dendropy in the necessary functions
- Lots of updates to the tests so that they actually work with the PICRUSt2-SC database and the split domains:
- tests/test_hsp.py: changed default file imports, split check sum tests to bacteria and archaea and added a chunk_size test
- tests/place_seqs.py: changed default file imports, split check sum tests to bacteria and archaea, change sepp test to run with archaeal reference files (smaller; previously this used the prokaryotic reference files, not the split domains), and split identify reference file tests for EPA-ng and SEPP to bacteria and archaea
- tests/test_pathway_pipeline.py: changed default file imports
- tests/test_data/pathway_pipeline files: EC: removed from numbers within files
- tests/test_data/add_descriptions files: EC: removed from numbers within files, ko: removed from files, remove COG and TIGRFAM that no longer exist, add test file for GO, made new expected output files to match new mapping file
- tests/test_util.py: changed default file imports, switched from COG and TIGRFAM tests to GO, added new md5sum values
- tests/test_data/workflow: EC: removed from files
tests/test_workflow.py: updated test_help_calls with the actual scripts that currently are used (i.e. removed _singleRef and oldIMG, added combine_domains.py and pick_best_domain.py).
All made on server and copied across to local.
- Other minor fixes to get Circle CI build to pass