Pathway activity inference

scRNA-seq yield many molecular readouts that are hard to interpret by themselves. One way of summarizing this information is by infering pathway activities from prior knowledge.

In this notebook we showcase how to use decoupler for pathway activity inference with the 3k PBMCs 10X data-set. The data consists of 3k PBMCs from a Healthy Donor and is freely available from 10x Genomics here from this webpage

Note

This tutorial assumes that you already know the basics of decoupler. Else, check out the Usage tutorial first.

Loading packages

First, we need to load the relevant packages, scanpy to handle scRNA-seq data and decoupler to use statistical methods.

[1]:
import scanpy as sc
import decoupler as dc

# Only needed for visualization:
import matplotlib.pyplot as plt
import seaborn as sns

Loading the data

We can download the data easily using scanpy:

[2]:
adata = sc.datasets.pbmc3k_processed()
adata
[2]:
AnnData object with n_obs × n_vars = 2638 × 1838
    obs: 'n_genes', 'percent_mito', 'n_counts', 'louvain'
    var: 'n_cells'
    uns: 'draw_graph', 'louvain', 'louvain_colors', 'neighbors', 'pca', 'rank_genes_groups'
    obsm: 'X_pca', 'X_tsne', 'X_umap', 'X_draw_graph_fr'
    varm: 'PCs'
    obsp: 'distances', 'connectivities'

We can visualize the different cell types in it:

[3]:
sc.pl.umap(adata, color='louvain')
../_images/notebooks_progeny_7_0.png

PROGENy model

PROGENy is a comprehensive resource containing a curated collection of pathways and their target genes, with weights for each interaction. For this example we will use the human weights (mouse is also available) and we will use the top 100 responsive genes ranked by p-value. To access it we can use decoupler.

[4]:
model = dc.get_progeny(organism='human', top=100)
model
[4]:
source target weight p_value
0 Androgen TMPRSS2 11.490631 0.000000e+00
1 Androgen NKX3-1 10.622551 2.242078e-44
2 Androgen MBOAT2 10.472733 4.624285e-44
3 Androgen KLK2 10.176186 1.944414e-40
4 Androgen SARG 11.386852 2.790209e-40
... ... ... ... ...
1395 p53 CCDC150 -3.174527 7.396252e-13
1396 p53 LCE1A 6.154823 8.475458e-13
1397 p53 TREM2 4.101937 9.739648e-13
1398 p53 GDF9 3.355741 1.087433e-12
1399 p53 NHLH2 2.201638 1.651582e-12

1400 rows × 4 columns

Activity inference with Multivariate Linear Model

To infer activities we will run the Multivariate Linear Model method (mlm), but we could do it with any of the other available methods in decoupler. It models the observed gene expression by using a regulatory adjacency matrix (target genes x pathways) as covariates of the linear model. The values of this matrix are the associated interaction weights. The obtained t-values of the fitted model are the activity scores.

To run decoupler methods, we need an input matrix (mat), an input prior knowledge network/resource (net), and the name of the columns of net that we want to use.

[5]:
dc.run_mlm(mat=adata, net=model, source='source', target='target', weight='weight', verbose=True)
58 features of mat are empty in 2635 samples, they will be ignored.
Running mlm on mat with 2638 samples and 13656 targets for 14 sources.
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:01<00:00,  1.90s/it]

The obtained scores (t-values)(mlm_estimate) and p-values (mlm_pvals) are stored in the .obsm key:

[6]:
adata.obsm['mlm_estimate']
[6]:
Androgen EGFR Estrogen Hypoxia JAK-STAT MAPK NFkB PI3K TGFb TNFa Trail VEGF WNT p53
AAACATACAACCAC-1 -0.254249 1.079791 -0.285906 -0.028353 -1.106584 -1.415070 -0.082578 -0.799087 -1.149092 0.683372 -0.574785 0.116901 0.123429 0.051064
AAACATTGAGCTAC-1 0.049356 1.810619 -0.642226 0.473845 0.044039 -2.655564 0.499077 -0.048742 -1.256668 -0.668830 0.662161 0.320902 -0.886260 -1.080493
AAACATTGATCAGC-1 -1.303848 0.994644 -1.424479 0.786483 0.760535 -1.422900 -0.092575 -0.283805 -0.821637 0.893210 0.002381 0.186219 -0.380405 -0.004127
AAACCGTGCTTCCG-1 -1.112302 0.619356 -0.535364 0.393064 5.351178 -1.143144 -1.695277 -0.320915 -0.656491 1.709201 -0.192290 -0.466818 -0.293868 -1.363734
AAACCGTGTATGCG-1 -0.024978 -0.952474 0.111260 0.429574 1.997764 1.380687 -0.279586 0.023845 -0.366084 -0.005061 0.697570 -0.186564 -1.324416 -0.745281
... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
TTTCGAACTCTCAT-1 -1.313532 1.637261 -0.521595 0.790193 5.909053 -1.263014 -2.046195 0.868877 -0.129425 2.279451 -0.395455 0.120465 0.379417 -0.726687
TTTCTACTGAGGCA-1 -0.089881 -0.277900 -0.336497 0.483560 1.307610 -0.530299 -0.635616 0.279144 -1.501107 0.174172 0.219880 0.288029 -0.357245 -0.364742
TTTCTACTTCCTCG-1 -0.384818 0.291916 -0.584801 -0.506899 0.398995 -0.762126 -1.212282 -1.737221 -0.691521 1.493315 1.031305 0.233905 -0.592073 -0.910929
TTTGCATGAGAGGC-1 -0.081163 1.160554 -0.053387 -0.944438 -1.008870 -1.162704 -0.632125 0.766052 -0.935325 0.303947 2.430733 -0.152342 0.621712 -0.182315
TTTGCATGCCTCAC-1 -0.215233 0.495030 0.697128 0.182926 1.069097 -0.965002 -0.787809 -1.315707 -0.759915 1.247356 0.310475 0.139115 0.036928 -1.155394

2638 rows × 14 columns

Note: Each run of run_mlm overwrites what is inside of mlm_estimate and mlm_pvals. if you want to run mlm with other resources and still keep the activities inside the same AnnData object, you can store the results in any other key in .obsm with different names, for example:

[7]:
adata.obsm['progeny_mlm_estimate'] = adata.obsm['mlm_estimate'].copy()
adata.obsm['progeny_mlm_pvals'] = adata.obsm['mlm_pvals'].copy()
adata
[7]:
AnnData object with n_obs × n_vars = 2638 × 1838
    obs: 'n_genes', 'percent_mito', 'n_counts', 'louvain'
    var: 'n_cells'
    uns: 'draw_graph', 'louvain', 'louvain_colors', 'neighbors', 'pca', 'rank_genes_groups'
    obsm: 'X_pca', 'X_tsne', 'X_umap', 'X_draw_graph_fr', 'mlm_estimate', 'mlm_pvals', 'progeny_mlm_estimate', 'progeny_mlm_pvals'
    varm: 'PCs'
    obsp: 'distances', 'connectivities'

Visualization

To visualize the obtained scores, we can re-use many of scanpy’s plotting functions. First though, we need to extract the activities from the adata object.

[8]:
acts = dc.get_acts(adata, obsm_key='mlm_estimate')
acts
[8]:
AnnData object with n_obs × n_vars = 2638 × 14
    obs: 'n_genes', 'percent_mito', 'n_counts', 'louvain'
    uns: 'draw_graph', 'louvain', 'louvain_colors', 'neighbors', 'pca', 'rank_genes_groups'
    obsm: 'X_pca', 'X_tsne', 'X_umap', 'X_draw_graph_fr', 'mlm_estimate', 'mlm_pvals', 'progeny_mlm_estimate', 'progeny_mlm_pvals'

dc.get_acts returns a new AnnData object which holds the obtained activities in its .X attribute, allowing us to re-use many scanpy functions, for example let’s visualise the Trail pathway:

[9]:
sc.pl.umap(acts, color='Trail', vcenter=0, cmap='coolwarm')
../_images/notebooks_progeny_19_0.png

It seem that in B and NK cells, the pathway Trail, associated with apoptosis, is more active.

Exploration

With decoupler we can also see what is the mean activity per group:

[10]:
mean_acts = dc.summarize_acts(acts, groupby='louvain', min_std=0)
mean_acts
[10]:
Androgen EGFR Estrogen Hypoxia JAK-STAT MAPK NFkB PI3K TGFb TNFa Trail VEGF WNT p53
B cells -0.468620 0.331232 -0.417349 -0.158265 0.814644 -0.850410 -0.588561 -0.209756 -0.882239 0.643023 1.154483 -0.204471 -0.220058 -0.571772
CD14+ Monocytes -0.757303 1.041441 -0.491022 0.747027 2.380898 -1.354828 -0.985024 -0.224320 -0.775338 1.474258 -0.269537 -0.301565 0.006913 -0.576483
CD4 T cells -0.739280 0.398468 -0.442663 0.255920 0.944010 -0.813422 -0.495894 -0.644337 -0.891932 0.913994 -0.273603 0.058538 -0.229640 -0.631877
CD8 T cells -0.729590 0.443396 -0.409557 0.302901 1.102024 -0.853103 -0.754845 -0.347034 -0.993045 1.228682 0.293810 0.072460 -0.133140 -0.677323
Dendritic cells -0.689356 1.080140 -0.639010 0.825671 3.072018 -1.557169 -0.519228 -0.528199 -0.769999 0.858208 -0.339825 -0.484119 0.182676 -0.613223
FCGR3A+ Monocytes -0.869172 1.303017 -0.634341 0.868262 3.638630 -1.599229 -1.155024 -0.105529 -1.160473 1.756404 -0.231644 -0.363514 -0.361645 -0.638332
Megakaryocytes -0.539583 2.361184 -0.379758 1.339222 0.074742 -2.248966 -0.628353 -0.999684 0.006636 0.659785 -0.047438 -0.015523 0.957471 -0.315496
NK cells -0.695226 0.543965 -0.211977 0.382988 2.203089 -0.974522 -1.161676 -0.094563 -1.086584 1.237731 0.729650 0.240034 -0.098274 -0.713052

We can visualize which group is more active using seaborn:

[11]:
sns.clustermap(mean_acts, xticklabels=mean_acts.columns, vmin=-2, vmax=2, cmap='coolwarm')
plt.show()
../_images/notebooks_progeny_24_0.png

In this specific example, we can observe that WNT to be more active in Megakaryocytes, and that Trail is more active in B and NK cells.