Table.
Comparison of Nanopore-based and Illumina-based drug resistance predictions using Mykrobe
| Number of false negatives (number of resistant isolates) | Number of false positives (number of Illumina susceptible isolates) | False negative rate | False positive rate | Positive predictive value | Negative predictive value | |
|---|---|---|---|---|---|---|
| Isoniazid | 0 (81) | 1 (70) | 0·0% (0·0–4·5) | 1·4% (0·3–7·7) | 98·8% (93·4–99·8) | 100·0% (94·7–100·0) |
| Rifampicin | 0 (79) | 0 (72) | 0·0% (0·0–4·6) | 0·0% (0·0–5·1) | 100·0% (95·4–100·0) | 100·0% (94·9–100·0) |
| Ethambutol | 0 (54) | 0 (97) | 0·0% (0·0–6·6) | 0·0% (0·0–3·8) | 100·0% (93·4–100·0) | 100·0% (96·2–100·0) |
| Pyrazinamide | 0 (30) | 0 (121) | 0·0% (0·0–11·4) | 0·0% (0·0–3·1) | 100·0% (88·6–100·0) | 100·0% (96·9–100·0) |
| Streptomycin | 0 (47) | 1 (104) | 0·0% (0·0–7·6) | 1·0% (0·2–5·2) | 97·9% (89·1–99·6) | 100·0% (96·4–100·0) |
| Amikacin | 0 (13) | 1 (138) | 0·0% (0·0–22·8) | 0·7% (0·1–4·0) | 92·9% (68·5–98·7) | 100·0% (97·3–100·0) |
| Capreomycin | 0 (13) | 1 (138) | 0·0% (0·0–22·8) | 0·7% (0·1–4·0) | 92·9% (68·5–98·7) | 100·0% (97·3–100·0) |
| Kanamycin | 0 (14) | 1 (137) | 0·0% (0·0–21·5) | 0·7% (0·1–4·0) | 93·3% (70·2–98·8) | 100·0% (97·3–100·0) |
| Ciprofloxacin | 0 (16) | 0 (135) | 0·0% (0·0–19·4) | 0·0% (0·0–2·8) | 100·0% (80·6–100·0) | 100·0% (97·2–100·0) |
| Moxifloxacin | 0 (16) | 0 (135) | 0·0% (0·0–19·4) | 0·0% (0·0–2·8) | 100·0% (80·6–100·0) | 100·0% (97·2–100·0) |
| Ofloxacin | 0 (17) | 0 (134) | 0·0% (0·0–18·4) | 0·0% (0·0–2·8) | 100·0% (81·6–100·0) | 100·0% (97·2–100·0) |
Data are % (95% CI) unless otherwise stated. For this comparison, we considered the Mykrobe resistance prediction from Illumina as the reference standard. A false negative means that Nanopore did not detect resistance but Illumina did. False positive means that Nanopore detected resistance but Illumina found susceptibility.