AllenCell / AllenCell/endothelial-pipeline
Remove instances of dynamically generating manifest names in favor of constants
- Vorherrschende Sprache
- Python
- Sterne
- 2
- Forks
- 0
- PR-Merge-Kennzahlen
- Keine gemergten PRs in 30 T.
Beschreibung
_Originally in #1641_
Something to discuss: I like the pattern of not dynamically generating any manifest names and instead having them defined in `settings` because it makes it easy to change names and is less error prone. All the outputs of the PCA workflows now use this pattern (see #1623) and I switched the bootstrapping manifests over to this pattern (see 7f50c61d39d01c1ac1c41821cfa6e130c7eec518), but not sure if we want to/how best to implement for the vector field/fixed point dataframes while still maintaining some flexibility with which features are included. Do you think it's likely at this point that we would change what features go into the 1D/2D/3D flow fields? Could/should we just simplify down to something like:
```python
COLUMNS_FOR_1D_FLOW_FIELD = [Column.DiffAEData.POLAR_ANGLE]
COLUMNS_FOR_2D_FLOW_FIELD = (
Column.DiffAEData.POLAR_RADIUS,
Column.DiffAEData.PC3_FLIPPED,
)
COLUMNS_FOR_3D_FLOW_FIELD = (
Column.DiffAEData.POLAR_ANGLE,
Column.DiffAEData.POLAR_RADIUS,
Column.DiffAEData.PC3_FLIPPED,
) # this is already defined as DYNAMICS_COLUMN_NAMES
```
and simplify manifest names to just `drift_vector_field_1/2/3D_grid.yaml` or do we need to maintain support for all "combinations" of 1D and 2D?
Beitragsleitfaden
Bewertung
Dieses Issue wurde noch nicht bewertet.