Table 1. Summary of synthetic data experiments.

In all cases, latent means and standard deviations were set as μ_j1 = 4, μ_j2 = 1, σ_j1 = 3 and σ_j2 = 2. Unless otherwise stated, our simulations were based on: I = 60 cells, J = 300 features, 20% ZI rate on average for the expression data (π_j = 0.20) and an average methylation coverage (n_ij) equal to 275 (sampled from a Uniform distribution with range [50, 500]) across cells and genes. When varying the number of cells, we use I ∈ {5, 10, 25, 50, 100, 200, 400, 800, 1600}. When varying expression ZI, we use π_j ∈ {0.1, 0.2, 0.3, 0.4, 0.5, 0.7, 0.8}. When varying methylation coverage, we sample n_ij from Uniform distributions with ranges given by [5, 10], [10, 20], [20, 50], [50, 250] and [500, 1000]. Full details are provided in S3 Text.

Experiment	Description
1	Correlations ρj sampled from a Beta(15, 15) distribution, varying number of cells.
2	Correlations ρj sampled from a Beta(15, 15) distribution, varying expression ZI.
3	Correlations ρj sampled from a Beta(15, 15) distribution, varying methylation coverage.
4	Correlations ρj sampled from a U[−0.8, −0.6] distribution, varying number of cells.
5	Correlations ρj sampled from a U[−0.8, −0.6] distribution, varying expression ZI.
6	Correlations ρj sampled from a U[−0.8, −0.6] distribution, varying methylation coverage.
7	As experiment 1, but latent expression means sampled from scVI.
8	As experiment 2, but latent expression means sampled from scVI.
9	As experiment 3, but latent expression means sampled from scVI.