nmf_models.plotting.factor_biplot

factor_biplot(nmf_model, TX=0, TY=1, n_labs=5, modality='rna', labs=None, select=None, title=None, x_lab=None, y_lab=None, mode='abs')[source]

Plot nmf embedding. With features projected on. Plot is scaled so that the projected features don’t exceed 1.

Parameters:

  • nmf_model (intNMF) – intNMF model with embed to view (.fit has been run)

  • TX (int) – factor/topic for x-axis

  • TY (int) – factor/topic for y-axis

  • n_labs (int) – number of features to plot per topic (i.e. 2)

  • modality (str, 'rna' | 'atac') – select modality from which features are projected from.

  • labs (list) – cell labels used to colour plot. Discrete or continuos determined by fraction of unique elements

  • select (list) – list of features to project

  • title (str) – plot title, default None

  • y_lab (str) – y-axis label, default None

  • x_lab (str) – x-axis label, default None

  • mode (str, 'abs' | 'diff') – select features to plot by absolute value or by difference

Returns:

  • matplotlib ax

  • list of features annotated