Table 1 Evaluation of small variant genotyping accuracy.

From: Targeted adaptive sampling enables clinical pharmacogenomics testing and genome-wide genotyping

DNA Input

N samples

Basecalling Model

Callability % (95% CI)

Concordance % (95% CI)

Sensitivity % (95% CI)

Specificity % (95% CI)

Precision % (95% CI)

1000ng

15

HAC

98.19 (97.89 - 98.50)

99.87 (99.80 - 99.94)

98.68 (97.75 - 99.61)

99.97 (99.94 - 99.99)

99.56 (99.21 - 99.92)

1000ng

15

SUP

99.57 (99.46 - 99.67)

99.90 (99.83 - 99.96)

98.94 (98.10 - 99.77)

99.97 (99.94 - 100.00)

99.63 (99.19 - 100.00)

800ng

9

HAC

97.90 (97.34 - 98.46)

99.91 (99.83 - 99.99)

99.64 (99.10 - 100.00)

99.92 (99.84 - 100.00)

98.91 (97.75 - 100.00)

800ng

9

SUP

99.38 (99.18 - 99.57)

99.90 (99.82 - 99.98)

99.65 (99.11 - 100.00)

99.91 (99.83 - 100.00)

98.76 (97.51 - 100.00)

All

24

HAC

98.08 (97.82 - 98.34)

99.89 (99.84 - 99.93)

99.04 (98.43 - 99.65)

99.95 (99.92 - 99.98)

99.32 (98.86 - 99.77)

All

24

SUP

99.49 (99.40 - 99.59)

99.90 (99.85 - 99.95)

99.20 (98.66 - 99.75)

99.95 (99.92 - 99.99)

99.30 (98.79 - 99.82)

  1. Performance metrics of the limit of detection (LOD) study evaluating small variant calling (SNPs and indels) across two DNA input levels (1000 ng and 800 ng) and two basecalling models (HAC and SUP). A total of 1,142 variants (1,103 SNPs and 39 indels) were analyzed across 31 pharmacogenes, excluding CYP2D6, CYP2A6, HLA-A, and HLA-B due to their high genomic complexity. Sequencing was performed on the Oxford Nanopore Technologies PromethION Pa2 Solo using a 3-plex sequencing strategy. Metrics shown include callability, concordance, sensitivity, specificity, and precision, each with corresponding 95% confidence intervals (CI).
Back to article page