GraphVelo 0.1.11

Inferring manifold-constrained velocity and transforming vectors across representations

GraphVelo allows for accurate inference of multimodal omics velocities and molecular mechanisms for single cells

GraphVelo is a graph-based machine learning procedure that uses RNA velocities inferred from existing methods as input and infers velocity vectors that lie in the tangent space of the low-dimensional manifold formed by the single-cell data.

Key Features

• Refine the velocity vectors estimated by any methods (e.g., splicing-based, metabolic labeling-based, pseudotime-based, lineage tracing-based, etc.) to the data manifold
• Infer modality dynamics that go beyond splicing events
  • Transcription rate of genes without introns or undergoing alternative splicing
  • Change rate of chromatin openness
  • More to be explored
• Serve as a plugin that can be seamlessly integrated into existing RNA velocity analysis pipelines
• Analyze dynamical systems in the context of multi-modal single-cell data

Getting Started with GraphVelo

Check the pipeline of RNA velocity estimation and you will find the niche of graphvelo:

Now let's get started with our Tutorials.

Installation

You need to have Python 3.8 or newer installed on your system.

To create and activate a new environment

conda create -n graphvelo python=3.8
conda activate graphvelo

Install via pip:

pip install graphvelo

Citing GraphVelo

Please see our manuscript for detailed explanation. If you find GraphVelo useful for your research, please consider citing our work as follows:

@article {Chen2024.12.03.626638,
	author = {Chen, Yuhao and Zhang, Yan and Gan, Jiaqi and Ni, Ke and Chen, Ming and Bahar, Ivet and Xing, Jianhua},
	title = {GraphVelo allows inference of multi-modal single cell velocities and molecular mechanisms},
	year = {2024},
	doi = {10.1101/2024.12.03.626638},
	publisher = {Cold Spring Harbor Laboratory},
	URL = {https://www.biorxiv.org/content/early/2024/12/07/2024.12.03.626638},
	journal = {bioRxiv}
}