FAQ

  • Q: What if I want to map only cell states, but not clones?
  • A: In theory, spaceTree can be used with any labels that can be transferred from scRNA-seq to spatial data. In practice, a way to achieve that at the moment requires a few modifications to the tutorial:

1) Currently the format of data object is fixed, but you can assign a dummy label to clone column in adata.obs by assigning roughly 50% of the cells to clone ‘0’ and the rest to ‘diploid’.

2) When you define the model, you need to set map_enteties to ‘type’ instead of ‘both’. E.g.:

model = GATLightningModule_sampler(data_param=data, weight_type=weight_type, map_enteties='type')

3) Instead of running get_results_calibrated(), you can use get_results_type(pred, data, node_encoder_rev, node_encoder_ct, activation='softmax') to get the results for cell types only.

 

  • Q: What if I want to map only clones, but not cell types?

  • A: In theory, spaceTree can be used with any type of label that can be transferred from scRNA-seq to spatial data. In practice, a way to achieve that at the moment requires a few modifications to the tutorial:

1) Currently the format of data object is fixed, but you can assign a dummy label to the ‘cell_type’ column in adata_ref.obs by assigning roughly 50% of the cells to dummy_cell_type1 and the rest to ‘dummy_cell_type2’ (or use any other names).

2) When you define the model, you must set map_enteties to ‘clone’ instead of ‘both’. E.g.:

model = GATLightningModule_sampler(data_param=data, weight_clone=weight_clone, map_enteties='clone')

3) Instead of running get_results_calibrated(), you can use get_results_clone(pred, data, node_encoder_rev, node_encoder_cl, activation='softmax') to get the results for clones only.

 

  • Q: What if I want to map a different type of label, e.g. cell types and cell cycle phases?

  • A: In theory, spaceTree can be used with any labels that can be transferred from scRNA-seq to spatial data. In practice, due to the fixed format of clonal labels, you might have to represent cell cycle phases as clones: e.g. “diploid”,”0”,”1”, etc.

  • Q: What if I want to use more than 2 sets of labels?
  • A: Right now we do not support more than 2 sets of labels. However, if you feel confident in your programming skills, you can modify the code of the model and the evaluation functions to support more than 2 sets of labels.