Table 1.
Summary statistics of genomic AST methods applied to simulated short-read metagenomes derived from data published in [10]. Performance metrics for metagenomic bins and isolate assemblies are aggregated over all short-read experiments (male and female E. coli mono-infection, K. pneumoniae/E. coli co-infection, K. pneumoniae/A. baumannii co-infection, and three-species co-infection) and 5 assayed antibiotic compounds. Metrics for resistomes are aggregated over resistome predictions as above and were scored against the true resistome as defined by the AST status of genomes used in the metagenome’s simulation. Values in brackets indicate 95% CI obtained by 1000× bootstrapping.
| Method | Data Type | bACC | ME | VME |
|---|---|---|---|---|
| WGS-AST (AREScloud) | isolate assembly | 0.956 (0.941–0.969) | 0.046 (0.032–0.061) | 0.042 (0.021–0.071) |
| WGS-AST (ARESdb) | metagenomic bins | 0.915 (0.912–0.919) | 0.057 (0.054–0.061) | 0.112 (0.106–0.119) |
| WGS-AST (ARESdb) | resistome (max) | 0.923 (0.917–0.928) | 0.136 (0.126–0.147) | 0.019 (0.016–0.022) |
| WGS-AST (ARESdb) | resistome (most certain) | 0.926 (0.922–0.931) | 0.023 (0.019–0.029) | 0.124 (0.116–0.131) |
| ResFinder 4 | isolate assembly | 0.811 (0.781–0.837) | 0.184 (0.156–0.213) | 0.195 (0.149–0.246) |
| ResFinder 4 | metagenomic bin | 0.711 (0.706–0.717) | 0.274 (0.267–0.28) | 0.303 (0.294–0.312) |
| ResFinder 4 | resistome (max) | 0.812 (0.804–0.819) | 0.284 (0.269–0.298) | 0.093 (0.086–0.1) |