scTransformPy

A Python implementation of of the scTransform method

Based on the R package sctransform originally by Christoph Hafemeister

• Hafemeister, C. & Satija, R. Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression. bioRxiv 576827 (2019). doi:10.1101/576827

Currently this supports basic functionality - variance stabilizing transform of UMI count data based on a general linear model and kernel-regularized parameters.

Installation

pip install git+https://github.com/deto/scTransformPy.git

Usage

import scTransform

residuals = scTransform.vst(
    umi, latent_var=['log_umi'], cell_attr
)

Feature comparison with original R package

Supported:

• cell_attr
• latent_var

Unsupported:

• batch variables (batch_var)
• non-regularized latent variables (latent_var_nonreg)
• gene sub-sampling (n_genes)
• cell sub-sampling (n_cells)
• alternate GLM fitting procedures (only the default, method="poisson" supported)
• sparse input umi matrix