Interactive stLearn

In this tutorial, we introduce several functions for interactive plotting to explore the progression of Ductal Carcinoma in situ (DCIS) - Invasive Ductal Cancer (IDC)

Source: https://support.10xgenomics.com/spatial-gene-expression/datasets

If you are using jupyter lab, then you need to run these two lines of code:

jupyter labextension install @jupyter-widgets/jupyterlab-manager
jupyter labextension install @bokeh/jupyter_bokeh

import stlearn as st
# Ingore all warnings
import warnings
warnings.filterwarnings("ignore")

### If encounter issues with Bokeh loading, below may be helpful ####
import os
os.environ["BOKEH_ALLOW_WS_ORIGIN"] = "localhost:8889"

We created a processed Visium data and you can now plot all interactive plots with stLearn.

#data = st.dataset.example_bcba() # TODO update this function
import scanpy as sc
data = sc.read_h5ad('/Volumes/GML001-Q1851/Brad/breast_LRCCIResults.h5ad')

data

AnnData object with n_obs × n_vars = 3813 × 20687
    obs: 'in_tissue', 'array_row', 'array_col', 'imagecol', 'imagerow', 'cell_type'
    var: 'gene_ids', 'feature_types', 'genome', 'n_cells'
    uns: 'cell_type', 'cell_type_colors', 'lr_cci_cell_type', 'lr_cci_raw_cell_type', 'lr_spot_indices', 'lr_summary', 'lrfeatures', 'per_lr_cci_cell_type', 'per_lr_cci_pvals_cell_type', 'per_lr_cci_raw_cell_type', 'spatial', 'spot_neighbours'
    obsm: '-log10(p_adjs)', 'lr_scores', 'lr_sig_scores', 'p_adjs', 'p_vals', 'spatial', 'spot_neighbours'

Interactive gene plot

We provide several options:

• Gene: Choose the available gene in the dataset

• Spot alpha: Choose the opacity value of spots

• Tissue alpha: Choose the opacity of H&E image

• Spot size: Choose the size of the spots

st.pl.gene_plot_interactive(data)

Interactive cluster plot with Pseudo-time-space results

We provide several options:

• Spot alpha: Choose the opacity value of spots

• Tissue alpha: Choose the opacity of H&E image

• Spot size: Choose the size of the spots

• Choose clusters: Choose clusters to be displayed

• Show spatial trajectories: Option to show spatial trajectories

st.pl.cluster_plot_interactive(data, use_label="louvain")

Ligand-receptor interactions plot

Displays the LRs score for significant spots.

Note that need to perform stLearn LR analysis first.

Options:

• Ligand receptor: Choose the ligand-receptor results to view; ranked from most no. of significant spots to least

• Tissue alpha: Choose the opacity of H&E image

• Spot size: Choose the size of the spots

st.pl.lr_plot_interactive(data)

Cell-cell interaction spatial plot

Displays significant Cell type-cell type interactions for selected ligand-receptor pairs in their spatial context.

Note that need to performed stlearn cci analysis first.

Options:

• Cell-type annotation select: Choose the cell type labels for which CCI permutation has been performed.

• Ligand receptor: Choose the ligand-receptor that facilitates the CCIs; ranked from most no. of significant spots to least

• Tissue alpha: Choose the opacity of H&E image

• Spot size: Choose the size of the spots

• Arrow size: The size of the arrows; direction indicates predicted direction of interaction (ligand->receptor)

st.pl.spatialcci_plot_interactive(data)