A Mixture Model Based Read Simulating Method for Single Cell RNA Sequencing

Techniques for single-cell RNA sequencing (scRNA-seq) has enabled unprecedented insights into gene expressions in cell level. Drop-seq is one of the prominent scRNA-seq protocols, and there has been a rapid growth in related analysis tools for Drop-seq data. These methods are tested either using spike-in experiments or on simulation datasets as the real word gene differential expressions are usually unknown. Since spike-in experiments are expensive and time consuming, simulated datasets have become a reasonable alternative method. However, current RNA-seq simulators mostly target at bulk RNA sequencing, which provokes the need of a scRNA-seq simulator for the Drop-seq technology. In this paper, we present a mixture model based read simulating method to simulate the sequencing reads of a Drop-seq experiment. The proposed method is able to simulate large amounts of Drop-seq reads according to the user's experimental setting. Data generated by the proposed model is a reasonable approximation to real Drop-seq data.