Landmarking
Landmarking is the process of assigning trusted cell-type labels to a small subset of cells in your dataset.
These landmark cell labels provide the initial supervision that CellTune uses before you can train any classifiers.
Each cell type should be represented. In practice, you need to label at least 10 cells per cell type (aim for 20 each) so that CellTune has enough examples to start learning.
Landmarking can be performed using either manual, automated/semi-automated supervised, or clustering-based approaches, or a combination of them.
Once you have landmark labels, you can proceed to the classification tutorial.