Table 2. Swift/T delivers on its promise of portability and scalability.
Synthetic data were generated using the NEAT synthetic read simulator [54]. Node sharing column indicates whether the cluster permits jobs to share the same node.
| System | Resource manager | Node type | # nodes per run | Node sharing | Test data |
|---|---|---|---|---|---|
| iForge [55] | PBS Torque | IvyBridge, 20 cores, 256 GB RAM |
1-8 | No | Soy NAM [56] using 2, 6, 12, or 16 sample batches † |
| XSEDE Stampede2 [57] | Slurm | KNL, 68 cores, 4 hardware threads/core, 96 GB DDR4, 16 GB MCDRAM |
1 | Yes | GIAB NA12878 Illumina HiSeq Exome (NIST7035) [58]; Synthetic chr1 exome seq 50X |
| Biocluster [50] | Slurm | Dell PowerEdge R620, 24 Cores, 384 GB RAM |
1; 3 | Yes | Synthetic WES 30X; Synthetic WES 50X; Synthetic WES 70X |
| Single server at CBSB, H3ABioNet node | N/A | HP Proliant dl380p gen. 8 24 cores 125 G RAM |
1 | Yes | Synthetic chr1 exome seq 50X |
† This Swift/T variant calling workflow was also used on iForge for a variety of analyses on WES and WGS data in different species.