memento is a Python package for estimating the mean, variability, and gene correlation from scRNA-seq data as well as contructing a framework for hypothesis testing of differences in these parameters between groups of cells. Method-of-moments estimators are used for parameter estimation, and efficient resampling is used to construct confidence intervals and establish statistical significance.

To install memento, pull the package from PyPI:

pip install -i https://test.pypi.org/simple/ memento

The most basic usage for memento is to test for differences in mean, variability, and coexpression between two groups of cells defined by the experiment. The following tutorials demonstrate quick usage cases:

• Analyzing the effect of exogenous IFN-B in CD14+ monocytes

memento is capable of handling experiments with multiple technical and biological replicates, such as batches/wells and different individuals respectively. The independent variable of interest can be defined at the cell level (environmental or genetic perturbation) or at a replicate level (SNPs in a population scale-study, individuals are replicates). The following tutorials demonstrate some more advanced use cases for memento:

• Performing mean eQTL analysis

• Simulations - Mean
  1. Run simulations/estimation_m/mean_estimation.py
  2. Generate plots with simulations/estimation_m/mean_comparison.ipynb notebook
• Simulations - Variance
  1. Generate simulation datasets with simluations/estimation_v/simulation_variance_datasets.py
  2. Run BASiCS with simluations/estimation_v/run_basics.r
  3. Run simluations/estimation_v/variance_estimation.py
  4. Generate plot with simulations/estimation_m/variance_comparison.ipynb notebook