celltometry

CellTOMetry: single-Cell Topologically Optimized Geometry

CellTOMetry is a python library to orchestrate topological single-cell data analysis. It is centered around TopOMetry and scanpy.

Installation and dependencies

CellTOMetry requires scanpy and TopOMetry. After installing both, install celltometry with:

pip3 install celltometry

Using CellTOMetry with scanpy

This is a quick-start. For further instructions, check TopOMetry documentation.

First, we load libraries and some data to work with:

import scanpy as sc
import topo as tp
import celltometry as ct

# Load the PBMC3k dataset
adata = sc.datasets.pbmc3k()

Next, we perform the default preprocessing workflow with scanpy: libraries are size-normalized, log-transformed for variance stabilization, and subset to highly variable genes.

# Normalize and find highly variable genes
adata = ct.preprocess(adata)

Then, we proceed to the default scanpy workflow. It corresponds to:

• Scaling data (optional, changes adata.X) - ``
• Performing PCA
• Learning a neighborhood graph
• Learn an UMAP projection with this graph
• Cluster this graph with the Leiden community detection algorithm

Similar to preprocessing, we wrap it with an one-liner:

adata = ct.default_workflow(adata, scale=True)

To run the topological workflow, create a TopOGraph object tg and use it to learn and add information to AnnData:

adata = ct.topological_workflow(adata, tg)

For further instructions, please check TopOMetry documentation.