EAGS: efficient and adaptive Gaussian smoothing applied to high-resolved spatial transcriptomics

Introduction

EAGS is a smoothing approach for spatial transcriptome data with ultrahigh resolution. Through the principal component information of the profiling of gene expression and the personally identifiable information between cells, it uses a two-factor adaptive smoothing weight mechanism to determine the neighborhood relationship of cells. Then it uses the self-developed smoothing weight formula to calculate the smoothing contribution and recalculate the gene expression of each cell.

This work belongs to the thematic series of Spatio-temporal omics algorithms.

Schematic Diagram

Figure: A. The spatio-temporal omics data processing flow. B. The EAGS method flow.

Quick Start

Download the GitHub Repository

Download this GitHub repository, and extract the contents into a folder.

Data Description

We provide a demo data data/demo.h5ad that generated by ScDesign3. The high-resolved transcriptomics data that we used can be downloaded from Zenodo (Mouse Olfactory Bulb: olfactory_bulb_correction_gmm.h5ad, Mouse Brain: SS200000141TL_B5_rawdata.h5ad, https://zenodo.org/record/7906815).

Option 1:Install and use it by Git

Install Dependencies

### Python enviroment constructed by Conda
conda create -n EAGS python=3.8
conda activate EAGS
pip install -r requirements.txt

Running EAGS Script from the Command-line

cd ./EAGS-main

# EAGS
python EAGS.py
--input data/demo.h5ad
--output data/demo_EAGS_smoothing.h5ad
--smooth_threshold 90
--a 1
--b 0
--n_comps 50
--n_neighbors 10
--normalize_zscore False

EAGS can also be performed directly without some parameter setting as:

# EAGS
cd ./EAGS-main

python EAGS.py
--input data/demo.h5ad
--output data/demo_EAGS_smoothing.h5ad

Option 2:Install and use it by Pypi

# Conda Terminal
pip install EAGS

# python Console
from EAGS import *
import scanpy as sc
adata = sc.read_h5ad('data/demo.h5ad')
result = gaussian_smooth_adaptively(adata = adata)

Parameter Explanation

The core function of the EAGS method is the gaussian_smooth_adaptively function provided in EAGS.py, and the detailed formula of this function has been given in the EAGS_function.py.

The following func is a quick example of running our Smoothing strategy. The function gaussian_smooth_adaptively() takes in anndata (H5AD format) object with the cell information (such as gene MID Counts in anndata.X , spatial coordinates in anndata.obsm['Spatial']). For more details on the input anndata (H5AD format), please check on link.

adata_after_smooth = gaussian_smooth_adaptively(adata=adata,
                                                smooth_threshold=90,
                                                a=1,
                                                b=0,
                                                n_comps=50,
                                                n_neighbors=10,
                                                normalize_zscore=True)

The parameters of gaussian_smooth_adaptively are:

• adata: The adata file (H5AD format) that need to be smoothed.
• smooth_threshold: Controling the gradient of Gaussian curve.
• a: The a-value of the Gaussian kernel function, which is used to control the height of the Gaussian curve.
• b: The b-value of the Gaussian kernel function, which is used to control the centre of the Gaussian curve.
• n_comps: Number of principal components for calculating neighborhoods.
• n_neighbors: Number of nearest neighbors for computing kernel bandwidth.
• normalize_zscore: Default preprocessing method for raw counts matrix, where "True" means the input anndata.X data that need to be normalized first and "False" means the input anndata.X data that have been normalized.

The output of gaussian_smooth_adaptively() is a anndata(H5AD format) includes:

• adata.X: MID counts that have been smoothed.
• adata.obsm[spatial]: Spot's spatial coordinates.

Tutorial

The analytical scripts used in this study and the corresponding results, various types of analysis and comparison methods are organized as Notebook (Turtorial/Turtorial_of_EAGS.ipynb).

Citation

If you use EAGS in your work, please cite it.

EAGS: efficient and adaptive Gaussian smoothing applied to high-resolved spatial transcriptomics

Tongxuan Lv, Ying Zhang, Mei Li, Qiang Kang, Shuangsang Fang, Yong Zhang, Susanne Brix, Xun Xu

GigaScience 2024 Feb 20. doi: 10.1093/gigascience/giad097.

Contact

Any questions or suggestions on EAGS are welcomed! Please report it on issues, or contact Tongxuan Lv ([email protected]) or Ying Zhang ([email protected]). We recommend using STOmics Cloud Platform to access and use it.