Table 3.
Prediction sensitivity for intragenus contamination for different QC metrics and genomic distance of contaminants (averaged over mixing ratio and species). Values are the number of contamination predictions (QC metrics above or below thresholds) [TP] divided by number of contaminated samples [P]. Good predictions (>0.8 accuracy) are colored in light blue, sufficient predictions (>0.2) in orange and insufficient predictions in white. Very good predictions (>0.99) are shown in bold.
| Predictor | Distant | Intermediate | Close |
|---|---|---|---|
| ConFindr cgMLST | 1.00 | 0.994 | 0.206 |
| ConFindr rMLST | 1.00 | 0.713 | 0.131 |
| # contigs | 0.87 | 0.838 | 0.438 |
| Dupl. BUSCO | 0.79 | 0.200 | 0.019 |
| Duplication ratio | 0.94 | 0.787 | 0.269 |
| GC content | 0.00 | 0.000 | 0.000 |
| Kraken2 contigs | 0.20 | 0.094 | 0.013 |
| Kraken2 reads | 0.25 | 0.000 | 0.000 |
| Duplicated mlst | 0.64 | 0.094 | 0.000 |
| N50 | 0.70 | 0.769 | 0.237 |
| Unique BUSCOs | 0.72 | 0.619 | 0.044 |
| assembly length | 0.80 | 0.450 | 0.119 |