General Tab
- You can select the different datasets you added in the "Data" Selection.
- With the "Color" Selection, you can select the column of your data, you want to be highlighted with color. The
selectable Options can be customized with the "data_variables" in main.py
- "OR" Filter: Filters all the selected Values, by connecting them with a logical "OR". For example all red OR blue
Objects. The selectable Options can be customized with the "data_variables" in main.py, and will show up with all
unique values of that data column in the selection.
- "AND" Filter: Filters all the selected Values, by connecting them with a logical "AND". For example all Objects, that
are red AND blue. The selectable Options can be customized with the "data_variables" in main.py, and will show up with
all unique values of that data column in the selection.
- "Export Data" Button exports the current view as .npy and .mat file in the output folder. If "Export only selection"
is enabled, only the data points currently on display will get exported (so for example all filtered data points won't
be included). The export file can be sorted by any sortable column.
Hyperparameter Optimization
PCA Preprocessing Tab
- The data can be Preprocessed via PCA. The reduction of PCA ("n_components") can be adjusted here.
- If you select "None" as Dimensionality Reduction, the PCA is restricted to 2-Dimensions.
Dimensional Reduction Parameters Tab
- Here you can change the Dimensional Reduction Method, as well as their parameters
- By default, these dimensional reduction methods are available:
- None (uses PCA as Dimensional Reduction Method)
- UMAP
- t-SNE
- PHATE
- CEBRA (in development)
Cluster Parameters
- "Update Cluster" Checkbox: When unchecking the box, the current clustering will be kept and not changed, when changing
the parameters (does not include change of dataset). (in development)
- Selection of other Cluster Algorithms (potential future feature)
- HDBSCAN used as clustering algorithm
- unclustered data points get assigned to -1
Cluster Selection Tab
- Lets you isolate a single cluster visually. Selected via entry of an integer or the slider (currently not functional)
Toolbar
- located on the right side of the plot
- General Tools
- Pan
- Box Zoom
- Wheel Zoom
- Save Plot
- Reset Plot Position
- Help
- Hover Tools
- Data point hover tool
- displays information of data point, when hovered over
- the information shown, can be customized with the "display_hover_variables" variable in main.py
- customized plot, based on data of data point (in development)
- Grid hover tool
- displays information about the quadrant and the data points in it, when hovered over it.
- customized plot, based on data of data points in quadrant (in development)
Grid Settings
- With this option you can enable a grid, separating your data points in quadrants, that can be hovered over with the
hover tool and displays you additional information
Statistics
- Some basic stats about the current selection
Other
- Introduces it's own data format and example to make your data compatible with DrCELL
- Buffering system to improve performance
- Dynamically generated plots for hover tool
- pip installation (experimental)
What's Changed
* Merging branches by lucakoe in https://github.com/lucakoe/DrCELL/pull/1
* Create python-package-conda.yml by lucakoe in https://github.com/lucakoe/DrCELL/pull/2
New Contributors
* lucakoe made their first contribution in https://github.com/lucakoe/DrCELL/pull/1
**Full Changelog**: https://github.com/lucakoe/DrCELL/commits/pre-release