Fig. 5: Performance evaluation of WarpDemuX on SQK-RNA004 for WDX4 (99.5% target accuracy, 95% yield) and WDX10 (99% target accuracy, 87% yield) models on Test Set 1 and 2 combined.
From: Demultiplexing and barcode-specific adaptive sampling for nanopore direct RNA sequencing
Details on the data sets are provided in Supplementary Table 14. Barcodes used per model are described in Supplementary Table 16. a, b Precision and recall metrics for individual barcodes in WDX4 and WDX10 configurations, respectively. c, d Receiver operating characteristic (ROC) curves for WDX4 and WDX10 configurations (black lines) vs. chance (diagonal gray dashed line). e, f Confusion matrices showing classification accuracy and cross-talk between barcodes for WDX4 and WDX10 sets. (g) Processing speed benchmarks on 8 CPU cores (11th Gen Intel(R) Core(TM) i7–1165G7, 2.80 GHz, 32GB RAM). Distribution of processing time per read (in min/100’000 reads) for 10 runs. Median shown in orange, the box indicates the interquartile range (IQR), and whiskers denote 1.5 × IQR with points outside the whiskers (flier points) shown. Source data are provided as a Source Data file.
