Transfer learning framework for single cell gene expression analysis in Python
Project description
scProject
Overview
Transfer learning for gene expression signatures (Python implementation of ProjectR).
Current version: 1.1.2 (PYPI)
Description
scProject is a software package that allows biologists to use transfer learning to understand the usage of previously discovered features in genomic datasets. Tools like NNMF allow biologists to find a set of vectors that linearly combine to approximately reconstruct a dataset. These features can have biological meaning as a pathway, cell type, gene program, cell feature, etc.
scProject leverages scikit-learn’s and scipy’s regression tools to effectively perform a change of basis with the new basis being the features discovered from NNMF or another previous analysis tool. For example, the values in the first column of the pattern matrix are the coefficients of the features to approximate the first sample in the dataset. This analysis allows the user to see the usage of features from a previously examined dataset and see if those same features are utilized in a new dataset. Using UMAP-learn to decrease the dimensionality of the regressed pattern matrix allows the user to visualize the pattern matrix in intuitive and informative ways.
Installation
scProject package can be installed by the following code:
pip install scProject
It is recommended to install the package and run the code in a virtual environment (venv). Use the code below to create and activate venv:
python -m venv env
source env/bin/activate
Contents
Additional Information
For more instructions and guidelines, read the documentation.
If you would like to leave a feedback, please contact cshin12@jhu.edu.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for scProject-1.1.2-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | a10cfb7c60da3f74eafefb241f7d6c4de5b14d5f96db52fe9f131de345b42b00 |
|
MD5 | aa90a7e850ced8b6cadbfb279f34396a |
|
BLAKE2b-256 | a4346ea64b57dc23bb5cc4a8ca24f58f0dd3f9b8aa1156bd21947df885f086f1 |