Fig. 3
From: Flexible and efficient handling of nanopore sequencing signal data with slow5tools
Merging, splitting, and querying BLOW5 files. a Normalized execution times for merging all individual batches of reads produced by MinKNOW into a single BLOW5 file for a typical ONT dataset (~9M reads), when using increasing numbers of parallel CPU threads. Data merging was evaluated on three separate machines with different disk systems (SSD, HDD, Distributed). For each system, execution times are normalized relative to the execution time for a single thread run on that system. Ideal curves (dashed lines) model the expected performance under optimal parallelism. b The same comparison for splitting a typical BLOW5 file into separate files for each read_group with slow5tools split. c The same comparison for retrieving a set of records from a BLOW5 file based on their read_ids with slow5tools get