Examples: Visium Breast cancer

The test data file locates at https://github.com/bioinfo-biols/RidgeSpace/tree/main/tests/VisiumCancer.h5ad.

[1]:

import RidgeSpace
import anndata as ad
import seaborn as sns
sns.set(style="ticks", font_scale=1.5)
adata = ad.read_h5ad('./tests/VisiumCancer.h5ad')#

Preproccess

Only need run once.

[2]:

RidgeSpace.tl_mesh(adata, mesh_optimization=True)

Radius   Count
274.116760    3308
274.614288    2491
273.748047    1372
274.000000      60
475.001038      41
dtype: int64
Use default radius 279.5990954589844
Triangles mesh saved in ./Ridge_tmp
Mesh optimization done

[3]:

RidgeSpace.tl_HE(adata, lightness=1.4, fast=False)#

Using V1_Breast_Cancer_Block_A_Section_1
HE information saved in ./Ridge_tmp

Denoising for each gene for ploting later. Only need run once.

[4]:

RidgeSpace.tl_denoise(adata, use_raw=False, plot_name = 'MGP')

Denoised value for MGP saved in ./Ridge_tmp

[5]:

RidgeSpace.tl_denoise(adata, use_raw=False, plot_name = 'AGR2')

Denoised value for AGR2 saved in ./Ridge_tmp

[6]:

RidgeSpace.tl_denoise(adata, use_raw=False, plot_name = 'FCGR3B')

Denoised value for FCGR3B saved in ./Ridge_tmp

Plotting

[8]:

RidgeSpace.pl_trajectory(adata, plot_name = 'MGP', obs_cluster='cluster', z_height=0.7, elev=45, view=210, \
                    zlim_min=0, lw=1, bg_alpha=0, normlabels=False, density_arrow=0.55, plot_HE=True, HE_z=0, return_ax=True)

[8]:

<Axes3DSubplot:xlabel='Spatial 1', ylabel='Spatial 2'>

[5]:

RidgeSpace.pl_trajectory(adata, plot_name = 'AGR2', obs_cluster='cluster', z_height=0.4, elev=40, view=250, \
                    zlim_min=0, lw=1, bg_alpha=0, density_arrow=0.6, min_length=0.1, plot_HE=True, HE_z=0, return_ax=True)

[5]:

<Axes3DSubplot:xlabel='Spatial 1', ylabel='Spatial 2'>

[6]:

RidgeSpace.pl_trajectory(adata, plot_name = 'pSpaceflow', obs_cluster='cluster', normalize_z=False, elev=60, view=190, lw=-0., \
                    zlim_min=0, return_ax=True, plot_HE=False, density_arrow=0.4, min_length=0.15)

No file named pSpaceflow in ./Ridge_tmp
Using noisy information in adata

[6]:

<Axes3DSubplot:xlabel='Spatial 1', ylabel='Spatial 2'>

[7]:

RidgeSpace.pl_trajectory(adata, plot_name = 'pseudotime', obs_cluster='cluster', normalize_z=False, elev=60, view=190, lw=-0., \
                    zlim_min=0, bg_alpha=-0, return_ax=True, plot_HE=False, density_arrow=0.5, min_length=0.15)

[7]:

<Axes3DSubplot:xlabel='Spatial 1', ylabel='Spatial 2'>