Skip to main content
. 2021 Feb 4;16(2):e0244471. doi: 10.1371/journal.pone.0244471

Fig 3. AWS WES benchmarks.

Fig 3

The runtime and cost for running the variant calling pipeline on a variety of AWS servers, ordered by increasing CPU/RAM resources. The GATK (Java/Intel) runs show minimal improvements in terms of runtime with more resources, whereas the elPrep (filters/sfm) runs scale nearly linearly. Because of its good scaling, the cost of using elPrep on more expensive servers with more resources is countered by the reduction in runtime.