Changelog for the ASCT+B Generator
- Added the code to verify the delimiter used in the input file, and to read the file accordingly based on the identified delimiter.
- Fixed alignment issues for the data with cell-independent features.
- Fixed alignment issues where the node is leaf and doesn't have any assigned cells or features.
- Fixed computing the depth of the anatomical structure.
- Added a feature to generate CSV files from TSV/CSV files.
- Added typings and docstrings.
- Modified the code according to PEP-8 style guide.
- Fixed a bug causing each reference to incorrectly span 5 columns. References now correctly span 3 columns.
- Added "note" and "ABBR" (abbreviation) fields for each feature. These columns are now required in the input file.
- The first 10 lines of the input file are now assumed to contain a descriptive header. This is duplicated to the output file. The 11th line in the input file is assumed to be a per-column header that is ignored.
- Test to be sure entities (anatomical structures, cell types, biomarkers and references) don't have commas in their names.
- Added a command line argument users can set to force anatomical structures to be unique (ie., each has one and only one parent). When set, if the same sub-structure exists in multiple parent-structures, then each child structure would need to be uniquely named. By default, a child structure can have multiple parent structures.
- Added a command line argument users can set to cause the program to automatically create missing features, when features are used. For example, if a biomarker is assigned to a cell type, by default, the biomarker must be independently defined but now users can optionally disable this requirement. Anatomical structures must always be defined.
-
Removed the (erroneous) assumption that an anatomical structure can only have a single parent, added more validation of the inputs, added debugging output options, and better handle command line arguments. Also various bug fixes.
-
These are the significant differences from version v0.1 and v1.0.
- The command line arguments have been greatly simplified.
- The number of AS levels is automatically computed.
- The input file has changed with this release. Cells are listed in a separate column from the children column.
- Biomarkers and references can now be added to any anatomical structure.
- The anytree Python module is required.
- A header is autogenerate in the output file.
- A DOT file can be generated to display the tree in Graphviz.
- Lots of tests to validate the input file
- This is a complete rewrite of the program. This version has none of the limitations from the alpha version, it includes more data validation, and requires less user intervention.
- This is a proof of concept and not meant for production use.