Fig. 1
A simple example to show the parameter effect or optimizer effect of NMI and ARI in scRNA-seq data on clustering. a This figure shows the relationship between mean gene expression levels and dropout rates. The black line indicates observed value, which is computed by the number of unexpressed cells divided by the number of cells; The red line represents expected value, which is calculated by negative binomial distribution with mean gene expression levels and dispersion parameter ψ(ψ=mean(ψi))b This figure shows how optimizers affect the performance of different methods on NMI and ARI. c-d These two figure indicate how the number of factors affect the NMI and ARI, respectively