Table 4 Clustering evaluation results. The NMI, AMI, and ARI scores are the cell type clustering results using SnapATAC software. “Real-input data” refers to the clustering results using the input real scATAC-seq data. Metrics for simATAC’s simulated bin-by-cell matrices for different noise levels are also compared presented for each of the Buenrostro2018, Cusanovich2018, and PBMCs datasets
| Mean | Standard deviation | NMI | AMI | ARI |
|---|---|---|---|---|
| Buenrostro2018 | ||||
| 0 | 0 | 0.71 | 0.83 | 0.54 |
| − 0.3 | 0.3 | 0.72 | 0.83 | 0.54 |
| − 0.4 | 0.4 | 0.67 | 0.78 | 0.50 |
| Real-input data | 0.54 | 0.62 | 0.36 | |
| Cusanovich2018 | ||||
| 0 | 0 | 0.96 | 0.97 | 0.94 |
| − 0.3 | 0.3 | 0.79 | 0.85 | 0.69 |
| − 0.4 | 0.4 | 0.62 | 0.71 | 0.40 |
| Real-input data | 0.41 | 0.46 | 0.22 | |
| PBMCs | ||||
| 0 | 0 | 0.78 | 0.87 | 0.71 |
| − 0.3 | 0.3 | 0.61 | 0.73 | 0.47 |
| − 0.4 | 0.4 | 0.56 | 0.67 | 0.44 |
| Real-input data | 0.45 | 0.54 | 0.28 | |