Table 1. NGSCheckMate performance.
| Data type | Dataset (pair type) | Sequencing depth1 | Individual | Sample | Test pairs #matched, #unmatched | Accuracy (%)2 |
|---|---|---|---|---|---|---|
| WGS (BAM) | TCGA colorectal (cancer versus normal) | >30X, down-sampling (0.5–30X) | 66 | 132 | 66, 66 | 100 |
| down-sampling (0.01–0.2X) | 55.3–99.2 | |||||
| WGS (BAM, FASTQ) | non-TCGA lymphoma (cancer versus normal) | 30–60X, down-sampling (0.5–30X) | 14 | 28 | 14, 28 | 100, 100 |
| WES (BAM) | TCGA 9 cancer types (cancer versus normal) | ∼100X | 421 | 842 | 421, 421 | 100 |
| TCGA kidney (cancer versus normal) | ∼100X, down-sampling (0.5–30X) | 50 | 100 | 50, 50 | 100 | |
| WES (FASTQ) | non-TCGA breast (cancer versus normal) | ∼60X, down-sampling (0.5–10X) | 68 | 136 | 68, 68 | 100 |
| RNA-seq (BAM) | TCGA colorectal (cancer versus normal) | ∼65X, down-sampling (0.5–10X) | 19 | 38 | 19, 19 | 100 |
| Single-cell WGS (BAM) | single-neuron | ∼42X, down-sampling (0.5–10X) | 3 | 36 | 210, 210 | 100 |
| glioblastoma (cancer–cancer) | 0.01–0.3X | 2 | 89 | 45, 45 | 87.8 | |
| Chip-seq (BAM, FASTQ) | within marks | 5.4 (2.2–19.0) | 8 | 119 | 72, 72 | 97.6, 97.7 |
| input versus mark | input DNA 2.3 (2.1–2.9) | 8 | 127 | 133, 133 | 98.5, 99.8 | |
| Panel-seq (BAM, FASTQ) | cancer versus normal, multiple regions, primary versus metastasis | 40 (20–119) | 5, 18, 11 | 12, 48, 25 | 92, 87 | 98.3, 99.4 |
| RNA-seq versus WES (BAM) | TCGA stomach (cancer or normal DNA versus cancer RNA) | RNA-seq (∼70X) WES (∼100X) | 65 | 201 | 132,132 | 100 |
| RNA-seq versus WES (FASTQ) | non-TCGA breast cancer (cancer or normal DNA versus cancer RNA) | RNA-seq (∼25X) WES (∼60X), down-sampling for both datasets (0.5–10X) | 53 | 159 | 106, 106 | 100 |
| RNA- versus ChIP-seq (BAM, FASTQ) | RNA-seq versus ChIP-seq (all marks) | RNA-seq (∼5X) ChIP-seq (described above) | 7 | 119 | 231, 231 | 99, 98.9 |
1For WGS, WES and RNA-seq, the average mean depth is shown. For Panel-seq and ChIP-seq, the average of the mean non-zero depths across the SNPs with at least one mapped read is shown with its range in parentheses.
2Accuracy estimates for the alignment-based and the alignment-free method are separated by a comma.