ladder.scripts.metrics.get_normalized_profile

ladder.scripts.metrics.get_normalized_profile#

ladder.scripts.metrics.get_normalized_profile(point_dataset, target=None, lib_size=10000.0, batched=False)#

Used to get library-size normalized pseudobulk profile from group of cells.

The normalized profile can then be compared to other profiles through RMSE or Pearson Correlation. Note that we cannot do this with the original data points if there is no explicit matching (as in the case of transfer).

Parameters:

  • point_dataset – The Dataset object to be provided.

  • target (Tensor, optional) – Optional target condition encoding tensor to subset the provided TensorDataset input.

  • lib_size (float, default: 1e4) – Library size to normalize each cell.

  • batched (bool, default: False) – If True, assumes batch is concatenated to the inputs

Returns:

normalized_profile : Tensor Normalized pseudo-bulk profile for the given points.