=========================================
- OrthoFinder will replace a few special characters within the protein ids with the underscore character, so this fix will make sure that the original protein ids can be retrieved.
- Major bug fix to handle possible missing links between features, mrnas, and cdss in genome object.
- Expansion of testing configuration to make it easier to iterate on a single large genome.
- Fix in table output where predicted orthologs were listed in multiple rows
-
Major release of kb_orthofinder
-
The reference genomes used to generate the protein families were updated to include the latest gene model versions released as part of Phytozome V12
-
"Annotate Plant Enzymes with OrthoFinder" has a new output table which captures the homologous relationships within the protein families and shows the user any putative orthologs that did not make the "cut"
-
Updated to use the Python 3 and minor bug fixes
-
- Major release of kb_orthofinder
- Includes bugfix for missing CDS fields
- Improved docs
- Added citations in PLOS format
- Module created by kb-sdk init