This project focuses on the analysis of single-cell RNA sequencing (scRNA-seq) data using various clustering methods. The primary objective is to identify and visualize clusters of cells based on their gene expression profiles.

A general workflow would involve importing the data, running some preprocessing and doing dimensionality reduction with PCA. Once that done, you can run clusterings.py and select different clustering methods and finally visualize the result by UMAP or tSNE.

In Jupyter Notebook, using the following command to install it.

!pip install git+https://github.com/ShiyuanWen/CSE185_FinalProject.git

Then, you can call the methods in the package after importing them.

from scRNA_clusterings import (
    kmeans_clustering, hierarchical_clustering, leiden_clustering,
    sc3s_clustering, benchmark_clustering, visualize_clustering
)

In local machine, using the following command to install it. The required packages are scanpy anndata matplotlib seaborn leidenalg sc3s scikit-learn

git clone https://github.com/ShiyuanWen/CSE185_FinalProject.git
cd CSE185_FinalProject
pip install -e .

Then, in your python script, call the methods in the package

from scRNA_clusterings import (
    kmeans_clustering, hierarchical_clustering, leiden_clustering,
    sc3s_clustering, benchmark_clustering, visualize_clustering
)

This project is designed to be an intermediate step in single cell analysis workflow. It provides various different clustering methods after performing dimensionality reduction. A general workflow would be similar to what we have done in CSE 185 lab6. For testing purpose, a harmony-processed PCA object can be used to test those functions.

In your python script, after performing PCA, apply

sc.pp.neighbors(adata_var)

and we can select different clustering methods such as kmeans_clustering, hierarchical_clustering, leiden_clustering, sc3s_clustering

Then, we can quantitatively analyze the quality of different clustering methods by calling benchmark_clustering

Finally, we can visualize the result in tSNE or UMAP by calling visualize_clustering

The scRNA_clusterings package provides several clustering methods for single-cell RNA sequencing data. Below are the available functions and their parameters.

Performs K-means clustering on the data.

Parameters:

• adata (AnnData): The annotated data matrix.
• n_clusters (int): The number of clusters.

Returns:

• np.ndarray: Cluster labels.

Performs hierarchical clustering on the data.

Parameters:

• adata (AnnData): The annotated data matrix.
• n_clusters (int): The number of clusters.

Returns:

• np.ndarray: Cluster labels.

Performs Leiden clustering on the data.

Parameters:

• adata (AnnData): The annotated data matrix.
• resolution (float, optional): The resolution parameter. Higher values lead to more clusters. Default is 1.0.

Returns:

• np.ndarray: Cluster labels.

Performs SC3S clustering on the data.

Parameters:

• adata (AnnData): The annotated data matrix.
• n_clusters (int): The number of clusters.

Returns:

• np.ndarray: Cluster labels.

Evaluates the clustering performance using silhouette scores.

Parameters:

• adata (AnnData): The annotated data matrix.
• labels (np.ndarray): Cluster labels.

Returns:

• float: Silhouette score.

Visualizes the clustering results using UMAP.

Parameters:

• adata (AnnData): The annotated data matrix.
• labels (np.ndarray): Cluster labels.
• title (str): Title for the plot.
• method (str, optional): Method for visualization. Can be 'umap' or 'tsne'. Default is 'umap'.

Returns:

• None: Displays the plot.