Fig. 11
From: A rank-based marker selection method for high throughput scRNA-seq data
Set up of the simulated data. We consider 3 conditions: all genes used for simulation, filtering after simulation, and filtering before simulation. On the left side of this diagramme, we produce 10 data sets by using all genes in simulation, and 10 more by filtering down to the 5000 most variable genes after simulation. These “filtering after simulation” data sets contain a subset of the information from the “all genes used for simulation” data sets. On the right hand side, we produce 10 data sets by filtering down to the 5000 most variable before simulation