Table 2.
Benchmark of ViReflow.
| # FASTQ pairs | Runtime (s) | Cost (USD) | Cost/Sample (USD) |
|---|---|---|---|
| 1S | 284 (4) | 0.01 (2 × 10–18) | 0.0100 |
| 10S | 255 (12) | 0.04 (0.003) | 0.0041 |
| 100S | 416 (21) | 0.49 (0.024) | 0.0049 |
| 1000S | 491 (9) | 5.65 (0.119) | 0.0057 |
| 10,000S | 12,075 (N/A) | 1197.53 (N/A) | 0.1198 |
| 684R | 8,267 (N/A) | 59.04 (N/A) | 0.0863 |
| 2607R,C | 4,144 (N/A) | 117.48 (N/A) | 0.0451 |
ViReflow was executed on 1, 10, 100, 1 K, and 10 K random 1000X depth sub-samplings of the single highest-depth sample from a NovaSeq SARS-CoV-2 amplicon sequencing run (denoted withS). ViReflow was also executed on two real NovaSeq runs (denoted withR), one of which was capped at 2 million successfully-mapped reads for each sample (denoted withC). All executions were run single-threaded. Total runtime (seconds) and total cost (US Dollars) across 10 technical replicates are shown as Mean (SD) pairs. “N/A” denotes single replicate execution due to high per-replicate compute costs. Specific details of tool choices (with versions) for each step of the pipeline can be found in the “Methods” section.