Splatter Simulation Of Single Cell Rna Sequencing

Splatter: simulation of single-cell RNA sequencing data

Splatter creates an interface for simulating RNA data. Their simulation has 2 steps: estimate parameters for sim from real dataset and then use those estimates to generate the synthetic dataset.

Simulation models

Their model is called Splat, but they also include other models to use as well.

Simple

This is a basic sim using negative binomial distribution.

Lun

This is build on Simple and adds scaling factors for each cell.

Lun 2

Exntension of the Lun model, and replace cell factors iwth library size factors, and batch effects that are applied to groups of cells in same batch.

scDD

checks differential expression between two groups of cells.

BASiCS

separates variation in scRNA-seq data into bio and technical components.

Splat

Their model. Captures features from real scRNA-Seq data. Usies parametric distirbutions with hyper-parameters based on real data estimates.

Conclusion

Splatter allows for simulations and their model Splat performs well, amongst the other models available on Splatter.