All API

mda.draw_triangle(prob, group, group_names, colors=array([[1, 0, 0], [0, 1, 0], [0, 0, 1]]), figsize=2, fs=12, **ka)

Draws triangular plot from LDA probabilities between 3 groups.

Parameters

  • prob (numpy.ndarray(shape=(n_cell,3))) – Probability of each cell being assigned to each group using coordinates on MDA.

  • group (numpy.ndarray(shape=(n_cell,))) – Group ID of each cell. Values must be 0,1,2, matching prob.shape[1].

  • group_names (List of str) – Names of groups.

  • colors (numpy.ndarray(shape=(3,3))) – Colors in [r,g,b] format for each group (as rows). 0<=r,g,b<=1

  • figsize (float) – Figure size (each dimension)

  • fs (float) – Font size

  • ka (dict) – Keyword arguments passed to seaborn.kdeplot

Returns

Figure drawn

Return type

matplotlib.pyplot.figure

mda.mda(data, group, datatype='raw', **ka)

Computes the Maximal Discriminating Axes (MDA) between cell groups in scRNA-seq.

Parameters

  • data (numpy.ndarray(shape=(n_gene,n_cell))) – Gene expression matrix. Can be raw read counts or log(CPM+1).

  • group (numpy.ndarray(shape=(n_cell,))) – Group ID of each cell. Each group must have at least 2 cells. Values must cover 0 to n_group-1.

  • datatype (str) –

    Type of data.

    • raw: Raw read counts

    • lcpm: Log(CPM+1). Natural log.

  • ka (dict) – Keyword arguments passed to sklearn.discriminant_analysis.LinearDiscriminantAnalysis

Returns

  • loc (numpy.ndarray(shape=(n_cell,n_group-1))) – Coordinates of each cell on the (n_group-1) dimensional MDA.

  • prob (numpy.ndarray(shape=(n_cell,n_group))) – Probability of each cell being assigned to each group using coordinates on MDA.

mda.sphere2tri(d)

Converts 3-dimensional probabilities to two-dimensional coordinates within unit triangle.

Uses stereographic projection from (0,0,-1), (0,-1,0), (-1,0,0), and then average.

Parameters

d (numpy.ndarray(shape=(3,n_cell))) – Input 3-dimensional probabilities

Returns

Two-dimensional coordinates within unit triangle at [0,0],[1,0],[0.5,sqrt(3)/2]

Return type

numpy.ndarray(shape=(2,n_cell))

mda.sphere2tri1(d)

Converts 3-dimensional sqrt probabilities to two-dimensional coordinates within unit triangle.

Uses stereographic projection from (0,0,-1).

Parameters

d (numpy.ndarray(shape=(3,n_cell))) – Input 3-dimensional probabilities

Returns

Two-dimensional coordinates within unit triangle at [0,0],[1,0],[0.5,sqrt(3)/2]

Return type

numpy.ndarray(shape=(2,n_cell))