Integrate a mosaic mouse brain dataset

In this notebook, we used SpaMosaic to integrate a postnatal mouse brain dataset. It has three sections where the first section consists of RNA and histone modification (H3K4me3) profiles while the second only has RNA expression and the third only has histone modification (H3K4me3) expression.

Data used in this notebook can be downloaded from google drive.

import os
import scanpy as sc
from os.path import join

from spamosaic.framework import SpaMosaic

os.environ['CUBLAS_WORKSPACE_CONFIG'] = ':4096:8'  # for CuBLAS operation and you have CUDA >= 10.2
import spamosaic.utils as utls
from spamosaic.preprocessing import RNA_preprocess, ADT_preprocess, Epigenome_preprocess

ad1_rna = sc.read_h5ad('./s1_adata_rna.h5ad')
ad1_epi = sc.read_h5ad('./s1_adata_H3K4me3.h5ad')
ad2_rna = sc.read_h5ad('./s2_adata_rna.h5ad')
ad3_epi = sc.read_h5ad('./s3_adata_H3K4me3.h5ad')

preprocessing

SpaMosaic expects batch-corrected low-dimensional representations for each modality as input. Therefore, for the raw count assays of each modality, SpaMosaic performs feature selection and dimension reduction to obtain the low-dimensional representations. Then, Harmony is performed to integrate these representations in each modality (optional, depending on the presence of strong batch effects).

In mosaic integration, SpaMosaic requires the input dataset in the following format:

{
    'rna':      [adata1_rna, adata2_rna,   None,         adata4_rna, ...],
    'protein':  [adata1_adt, None,         adata3_adt,   None,       ...],
    'atac':     [None,       adata2_atac,  None,         None,       ...],
    'histone':  [None,       None      ,   adata3_hist,  None,       ...],
    ...
}

In the dictionary, each key represents a modality and each modality key corresponds to list of anndata objects. Each anndata object contains modality-specific information for a particular section. For example, the first object adata1_rna under the ‘rna’ key holds the RNA profiles for the first section, while the first object adata1_adt under the ‘protein’ key holds protein profiles for the same section. If a section is not measured for a particular modality, its value in the list is None. For instance, the first element under the ‘atac’ and ‘histone’ keys is None, indicating that the first section was not measured with ATAC or histone modality. All lists have the same length, which corresponds to the number of sections in the target dataset.

SpaMosaic employs contrastive learning to capture the relationships between modalities. To achieve this, it requires the modalities in a mosaic dataset to be directly or indirectly connected through one or multiple sections. If a pair of modalities occur in the same section, there is a direct connection between this pair of modalities, while an indirect connection requires multiple intermediary direct connections.

# for this dataset, we have
input_dict = {
    'rna':     [ad1_rna, ad2_rna, None],
    'H3K4me3': [ad1_epi, None,    ad3_epi]
}

input_key = 'dimred_bc'

ad1_rna

AnnData object with n_obs × n_vars = 9548 × 21536
    obs: 'src'
    obsm: 'spatial'
    layers: 'counts'

Parameters in the following RNA_preprocess and Epigenome_preprocess function:

• batch_corr: whether to perform batch correction with Harmony

• n_hvg/n_peak: how many highly variable genes/peaks to select

• batch_key: the key in the row metadata that holds the batch labels

• key: the key in .obsm that will hold the output low-dimensional representations

After preprocessing, the low-dimensional representations can be accessed by: ad1_rna.obsm[input_key], ad1_epi.obsm[input_key], ad2_rna.obsm[input_key]…

RNA_preprocess(input_dict['rna'], batch_corr=True, n_hvg=5000, batch_key='src', key=input_key)
Epigenome_preprocess(input_dict['H3K4me3'], batch_corr=True, n_peak=50000, batch_key='src', key=input_key)

Use GPU mode.
        Initialization is completed.
        Completed 1 / 10 iteration(s).
        Completed 2 / 10 iteration(s).
        Completed 3 / 10 iteration(s).
Reach convergence after 3 iteration(s).
Use GPU mode.
        Initialization is completed.
        Completed 1 / 10 iteration(s).
        Completed 2 / 10 iteration(s).
Reach convergence after 2 iteration(s).

training

SpaMosaic employs modality-specific graph neural networks to embed each modality’s input into latent space. In horizontal integration, all sections have only one modality, thus each section has only one set of embeddings.

The crucial parameters include:

• intra_knns: how many nearest neighbors to consider when searching for spatial neighbors within each section (list or integer)

• inter_knn_base: how many nearest neighbors to consider when searching for mutual nearest neighbors between sections (integer)

• w_g: the weight for spatial-adjacency graph

The following parameters are recommended to use in complex integration scenarios, like varying resolution or size across sections

• smooth_input: whethere to smooth the input representations (bool)

• smooth_L: number of LGCN layers for smoothing input

• inter_auto_knn: whether to automatically balance the kNN during MNN searching (bool)

• rmv_outlier: whether to remove outlier of MNN (bool)

• contamination: percentage of removed MNN outlier

for training:

• net: which graph neural network to use (only support wlgcn now)

• lr: learning rate

• n_epochs: number of training epochs

• w_rec_g: the loss weight for reconstructing original graph structure. If the target dataset contains protein modality, we recommend a low value for w_rec_g (e.g., 0); if it contains epigenome modality, we recommend a high value for w_rec_g (e.g., 1).

model = SpaMosaic(
    modBatch_dict=input_dict, input_key=input_key,
    batch_key='src', intra_knns=10, inter_knn_base=10, w_g=0.8,
    seed=1234,
    device='cuda:0'
)

model.train(net='wlgcn', lr=0.01, T=0.01, n_epochs=100)

batch0: ['rna', 'H3K4me3']
batch1: ['rna']
batch2: ['H3K4me3']
------Calculating spatial graph...
The graph contains 95480 edges, 9548 cells.
10.0000 neighbors per cell on average.
------Calculating spatial graph...
The graph contains 93700 edges, 9370 cells.
10.0000 neighbors per cell on average.
------Calculating spatial graph...
The graph contains 95480 edges, 9548 cells.
10.0000 neighbors per cell on average.
------Calculating spatial graph...
The graph contains 24990 edges, 2499 cells.
10.0000 neighbors per cell on average.
Searching MNN within rna
Finding MNN between (mult, rna) using KNN (10, 10)
Number of mnn pairs for rna:12129
Searching MNN within H3K4me3
Finding MNN between (mult, atac) using KNN (10, 10)
Number of mnn pairs for H3K4me3:8201

100%|███████████████████████████████████████████████████████████████████████████████████| 100/100 [00:05<00:00, 18.35it/s]

inference

After training, SpaMosaic can infer the modality-specific embedding for each section. These embeddings will be directly saved in original anndata objects. For example, the RNA-specific embeddings can be accessed by ad1_rna.obsm['emb'], ad2_rna.obsm['emb'], the histone modification-specific embeddings can be accessed by ad1_epi.obsm['emb'], ad3_epi.obsm['emb'].

The following infer_emb function will return a new list of anndata objects, which save the final/merged embeddings for spatial clustering. The final embeddings can be accessed by ad_embs[0].obsm['merged_emb'], ad_embs[1].obsm['merged_emb'], …

In mosaic integration, the final merged embeddings are obtained by averaging the modality-specific embeddings.

ad_embs = model.infer_emb(input_dict, emb_key='emb', final_latent_key='merged_emb')
ad_mosaic = sc.concat(ad_embs)
ad_mosaic = utls.get_umap(ad_mosaic, use_reps=['merged_emb'])

/home/yanxh/anaconda3/envs/spamosaic-env/lib/python3.8/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from .autonotebook import tqdm as notebook_tqdm

Visualize the final embeddings using UMAP plots, spot are colored by batch labels

utls.plot_basis(ad_mosaic, basis='merged_emb_umap', color=['src'])

clustering

Parameters for the following utls.clustering function:

• n_cluster: expected number of clusters

• used_obsm: the target key in the ad_mosaic.obsm that holds the embeddings for clustering input

• algo: which clustering algorithm to use, mclust or kmeans. Sometimes mclust can fail to output clustering, it will automatically output the kmeans outputs.

• key: the key in the row metadata that will hold the output clustering labels.

Generally, SpaMosaic works better with the mclust algorithm in spatial domain identification task. The final/merged embeddings are used as input for mclust and the clustering labels can be accessed by ad_mosaic.obs['mclust'].

utls.clustering(ad_mosaic, n_cluster=7, used_obsm='merged_emb', algo='mclust', key='mclust')
utls.split_adata_ob(ad_embs, ad_mosaic, 'obs', 'mclust')

R[write to console]:                    __           __
   ____ ___  _____/ /_  _______/ /_
  / __ `__ \/ ___/ / / / / ___/ __/
 / / / / / / /__/ / /_/ (__  ) /_
/_/ /_/ /_/\___/_/\__,_/____/\__/   version 6.0.0
Type 'citation("mclust")' for citing this R package in publications.

fitting ...
  |======================================================================| 100%

utls.plot_basis(ad_mosaic, basis='merged_emb_umap', color=['src', 'mclust'])

Spatial plots of clustering label for individual sections

sizes = [30, 30, 100]
for i, ad in enumerate(ad_embs):
    utls.plot_basis(ad, 'spatial', 'mclust', s=sizes[i])