Skip to main content

Table 1 Computational performance of each tool

From: Gramtools enables multiscale variation analysis with genome graphs

 

Index

Map and genotype

 

Disk (Mb)

Max RAM (Mb)

Speed (sec)

Max RAM (Mb)

Speed (reads/sec)

vg

29

609

105

605[158]

3,961

gramtools

153

480

32

632

34,290

GraphTyper2

869[88]

7,604

  1. Index: genome graph processing step allowing subsequent read mapping. For vg, includes a graph pruning step to reduce graph complexity (else temporary disk use exceeded 500 Gb before completion, see “vg and GraphtTyper2 genome graph construction” section). GraphTyper2 has no separate indexing operation. Map and Genotype: Speed shows the average number of reads mapped across the 17 samples (10.7 million) divided by the average CPU time. vg and GraphTyper2 have separate read mapping and genotyping steps: for speed, CPU time is summed, and for RAM, mapping is shown followed by genotyping in square brackets. GraphTyper2 does not implement its own mapping but requires an input file of reads mapped to a linear reference genome; mapping RAM and speed is shown for bowtie2 with default parameters. metrics: Mb, megabytes; sec, total CPU seconds (accounts for multi-threading, 10 threads used for genotyping in each tool)