Skip to main content
. 2021 Apr 19;17(4):e1008926. doi: 10.1371/journal.pcbi.1008926

Table 8. Benchmarking SmartMap software.

Pre-algorithm alignment and processing Read allocation algorithm
Read Alignment Processing Alignments Reading alignments
Algorithm CPU Time Wall Time CPU Time Proc. File Size CPU Time Max. Memory Algorithm Time Avg. MAE
SmartMap 317:30:25 6:39:46 1:34:29 59 GB 0:16:49 53 GB 0:42:38 4.04
BM-Map 317:30:25 6:39:46 N/A 820 GB 6:25:09 146 GB ERROR
Iteration 0 317:30:25 6:39:46 1:34:29 59 GB 0:16:10 53 GB 0:35:16 4.12
Random 317:30:25 6:39:46 2:15:12 15 GB 0:03:58 39 GB 0:15:46 5.48
Uniread 36:08:04 0:45:34 0:17:07 13 GB 0:03:19 39 GB 0:14:43 6.50

Benchmarking conducted on computer with Ubuntu 20.04.1 LTS with 224 GB of RAM and dual Intel Xeon CPU E5-2690 v3 @ 2.60GHz processors, running on one thread except the read alignment, which used 48 threads. All times are represented in hours:minutes:seconds.

Alignment conditions are identical for all but Uniread. Parsing reads is typically conducted in parallel with alignment. File size represents the size of the required file after read parsing needed for the algorithm in question. Reading alignments is part of each algorithm and is included in the Algorithm Total Time.

Average Mean Absolute Error (Avg. MAE) is computed against the gold standard on the set of true origin loci. These benchmarking analyses were conducted separately with separate alignments from the analyses in Fig 3, and the avg. MAEs vary slightly in magnitude from those presented in Fig 3.