Skip to main content

Table 2 MiniScrub improves read error correction in the metagenome setting

From: De novo Nanopore read quality improvement using deep learning

  No scrubbing With scrubbing
Avg. coverage pct. 47.04% 52.71%
Avg. mean coverage depth 3.08 3.34
Pct. of genomes above 1.0 coverage depth 46.67% 60.00%
  1. Reads from the high complexity (HC) dataset [31], both with and without scrubbing beforehand, were corrected using Canu’s [19] error correction module and then aligned to their reference genomes with GraphMap [26]. The statistics in the table are averages across all source genomes that had non-zero coverage. Applying scrubbing before read error correction improves average coverage percentage and the average mean coverage depth across the source genomes, and leads to a larger number of source genomes having a mean coverage depth of at least 1.0 Best performance numbers are shown in bold