Simulate single-cell RNA-SEQ data using the Splatter statistical framework, which is described here but implemented in python. In addition, simulates doublet cells and cells with shared gene-expression programs. This was used to benchmark methods for gene expression program inference in single-cell rna-seq data as described here
run_scsim.py has example code for running a simulation with a given set of parameters. It saves the results in the numpy compressed matrix format which can be loaded into a Pandas dataframe as follows:
with np.load(filename) as f:
result = pd.DataFrame(**f)