Currently, we have the option of adding two matrices:

1. Test matrix (expression_data)
2. Data transformation (normalisation, log, etc), which accepts a new matrix file.

If we want to create a new column based on a feature, say FOXP3 which will contain FOXP3p and FOXP3n, it will use (1). We want to be able to use (2) in some cases. However, (2) is loaded after selecting cells. So we would need to fix that first. So...

1. Loading (2) matrix so it's available for the filtering step.
2. Choose where the data transformation will take place. Especially needed for the filtering step.
   • We need this for the summary statistics and feature filtering
   • Exclude log transformations, which is done only until the visualization step.
3. We can't avoid normalizing all cells before filtering if we want to use (2) to create a new column.

Perhaps the easiest solution is creating a new metadata file with the column... but we are a bit lazy and want to save space.