Analysis of 10x Visium human breast cancer slice#

In this tutorial, we demonstrate SpaSRL on the analysis of 10x Visium human breast cancer (block A section 1) slice including

Spatial expression enhancement
Self-representation learning
Spatial domain identification
Spatial domain annotation
Finding differentially expressed genes

The dataset is available at 10x genomics website (Spatial Gene Expression >> Visium Demonstration (v1 Chemistry) >> Space Ranger 1.0.0 >> Human Breast Cancer (Block A Section 1)).

[1]:

import numpy as np
import pandas as pd
import scanpy as sc
import matplotlib.pyplot as plt
import SpaSRL

%matplotlib inline

Data loading and preprocessing#

We load the dataset and find top 2000 highly variable genes.

[2]:

adata = sc.read_visium('./data/Human Breast Cancer (Block A Section 1)/')
adata.var_names_make_unique()
adata

Variable names are not unique. To make them unique, call `.var_names_make_unique`.
Variable names are not unique. To make them unique, call `.var_names_make_unique`.

[2]:

AnnData object with n_obs × n_vars = 3813 × 33538
    obs: 'in_tissue', 'array_row', 'array_col'
    var: 'gene_ids', 'feature_types', 'genome'
    uns: 'spatial'
    obsm: 'spatial'

[3]:

sc.pp.highly_variable_genes(adata, n_top_genes=2000, flavor='seurat_v3')

Spatial expression enhancement#

We perform spatial expression enhancement to aggregate expression from spatial neighbors.

[4]:

SpaSRL.spatial_enhancement(adata)

Self-representation learning#

We perform self-representation learning on the enhanced data.

[5]:

SpaSRL.run_SRL(adata, n_neighbors=30, n_pcs=15)

 21%|██████▏                       | 104/500 [00:09<00:37, 10.62it/s, relChg: 3.851e-05, recErr: 9.747e-06, converged!]

Spatial domain identification#

We identify spatial domians using the representation matrix.

[6]:

sc.tl.leiden(adata, resolution=0.65, neighbors_key='representation')
SpaSRL.refine_spatial_domains(adata, domain_key='leiden')

[7]:

fig, axs = plt.subplots(figsize=(8, 8))

sc.pl.spatial(
    adata,
    img_key='hires',
    color='leiden_refined',
    size=1.5,
    palette=sc.pl.palettes.default_102,
    legend_loc='right margin',
    show=False,
    ax=axs,
)

plt.tight_layout()

C:\ProgramData\Anaconda3\lib\site-packages\anndata\_core\anndata.py:1220: FutureWarning: The `inplace` parameter in pandas.Categorical.reorder_categories is deprecated and will be removed in a future version. Reordering categories will always return a new Categorical object.
  c.reorder_categories(natsorted(c.categories), inplace=True)
... storing 'feature_types' as categorical
C:\ProgramData\Anaconda3\lib\site-packages\anndata\_core\anndata.py:1220: FutureWarning: The `inplace` parameter in pandas.Categorical.reorder_categories is deprecated and will be removed in a future version. Reordering categories will always return a new Categorical object.
  c.reorder_categories(natsorted(c.categories), inplace=True)
... storing 'genome' as categorical

../_images/tutorials_10x_Visium_breast_cancer_16_1.png

Spatial domain annotation#

We annotate the spatial domains based on the pathologist annotation from Fu, H. et al.

[8]:

map_dict = {
    '0': 'Surrounding tumor',
    '1': 'Invasive',
    '2': 'Invasive',
    '3': 'Healthy',
    '4': 'Invasive',
    '5': 'Tumor',
    '6': 'Tumor',
    '7': 'Tumor',
    '8': 'Invasive',
    '9': 'Surrounding tumor',
}

[9]:

adata.obs['annotation'] = pd.Categorical(
    adata.obs['leiden_refined'].map(map_dict),
    categories=['Invasive', 'Tumor', 'Surrounding tumor', 'Healthy']
)

[10]:

fig, axs = plt.subplots(figsize=(8, 8))

sc.pl.spatial(
    adata,
    img_key='hires',
    color='annotation',
    size=1.5,
    palette=sc.pl.palettes.default_102,
    legend_loc='right margin',
    show=False,
    ax=axs,
)

plt.tight_layout()

../_images/tutorials_10x_Visium_breast_cancer_21_0.png

Finding differentially expressed genes#

We find the differentially expressed (DE) genes across identified domains and show their expression patterns in spatial coordinates.

[11]:

sc.tl.rank_genes_groups(adata, groupby='leiden_refined', use_raw=False, layer='counts', method='t-test')

[12]:

DE_genes = pd.DataFrame(adata.uns['rank_genes_groups']['names']).iloc[:10,:]
DE_genes

[12]:

	0	1	2	3	4	5	6	7	8	9
0	IGLC2	CXCL14	COX6C	ACKR1	CRISP3	CPB1	MGP	S100G	LINC00645	ALB
1	IGHG1	CCND1	SLC39A6	CCL21	SLITRK6	HLA-B	S100G	HLA-B	MUC5B	MGP
2	IGLC3	DEGS1	SNCG	CLDN5	S100A13	IL6ST	DSP	HLA-A	IGFBP2	PGC
3	IGKC	CPNE7	WFDC2	CHRDL1	SERHL2	ADIRF	TFF3	AC087379.2	PVALB	MT-ATP8
4	C3	DEGS2	MT-ND1	MMRN1	IGFBP5	HLA-C	MT-CO2	IFI6	ZNF703	HBA2
5	IGHG3	PSMB4	CSTA	ADAM33	CALML5	CFB	ACADSB	IFI27	SLC39A6	MT-ND2
6	SFRP4	MUC1	MCCD1	TNXB	S100A16	HLA-A	MT-CO1	IFT122	EXOC2	S100A7
7	JCHAIN	S100A11	MT-CO1	TTN	KRT8	COX6C	TFF1	BST2	COLEC12	HBB
8	IGHM	GFRA1	RAB11FIP1	INMT	S100A11	H2AFJ	IGFBP5	ISG15	SLC30A8	CCL7
9	IGHG2	AGR2	UGCG	CCL19	NUPR1	FCGRT	LDHA	LGALS3BP	PLXNB2	STEAP4

[13]:

fig, axs = plt.subplots(figsize=(8, 8))

sc.pl.spatial(
    adata,
    img_key='hires',
    color=DE_genes.iloc[0,0],
    layer='log1p',
    size=1.5,
    cmap='summer',
    vmin='p50',
    vmax='p99',
    show=False,
    ax=axs,
)

plt.tight_layout()

../_images/tutorials_10x_Visium_breast_cancer_26_0.png

[14]:

fig, axs = plt.subplots(figsize=(8, 8))

sc.pl.spatial(
    adata,
    img_key='hires',
    color=DE_genes.iloc[0,1],
    layer='log1p',
    size=1.5,
    cmap='summer',
    vmin='p50',
    vmax='p99',
    show=False,
    ax=axs,
)

plt.tight_layout()

../_images/tutorials_10x_Visium_breast_cancer_27_0.png