Albacore version 2

HDF5 raw data were acquired by MinKNOW version 1.7 (Oxford Nanopore Technologies, UK), and local base-calling with demultiplexing setting was performed using Albacore version 2.3.4 (Oxford Nanopore Technologies, UK). The quality of the sequencing raw data was assessed using NanoStat^{127 (link)}. The data was indexed using Nanopolish^{128 (link)} to link each sequencing read with its signal-level data in the HDF5 files. The base-called and indexed sequencing reads were aligned to the human reference genome assembly GRCh38 using BWA-MEM¹²⁹ with the ont2d mode. SAMtools^{130 (link)} was used to sort and index the aligned sequencing reads prior to variant- and methylation-calling via Nanopolish. The pipelines for the variant- and methylation-calling are described in more detail in the Supplementary Notes. The functional effect of the variants were predicted using open-source algorithms including PolyPhen-2^{131 (link)}, PANTHER^{132 (link)}, Envision¹³³ , MutationAssessor^{134 (link)}, MutPred2¹³⁵ and SNPs&GO^{136 (link)}. The mtDNA deletion was also determined via MitoDel¹³⁷ and eKLIPse^{138 (link)}. MitoDel is a tool for detecting and quantifying mtDNA deletions even at low heteroplasmy levels via the BLAT split read mapping method. The eKLIPse pipeline uses soft clipping alignment analysis of sequencing reads to predict mtDNA deletions.

HMW DNA sequencing libraries were prepared from 400 ng of input DNA using the Ligation Sequencing kit (SQK-LSK109) along with the Native Barcoding Expansion Kit (EXP-NBD104) following the manufacturer’s (Oxford Nanopore Technologies, UK) protocols. The final libraries were then sequenced on a single flow cell (FLO-MIN106D) on the MinION, which was controlled using the MinKNOW version 3.4.8 software. Real-time base calling was turned off and was instead performed on the servers of the CSC—IT Center for Science, Finland, using Albacore version 2.3.4 (Oxford Nanopore Technologies, UK).

MinION sequencing was performed using the MinION Mk1b sequencer and FLO-MIN106 flow cells. Nucleotides of each read were called by Albacore version 2.1.3 (Oxford Nanopore Technologies), and the sequences were deposited in the DDBJ DRA database (https://www.ddbj.nig.ac.jp/dra/index-e.html) under the accession numbers DRR187692 to DRR187701. Bacterial species were assigned using the minimap2 software^{15 (link)} with the bacterial genomes obtained from the GenomeSync database (http://genomesync.org) as we previously reported^{12 (link),13 (link)}.
We defined the VBD index as the proportion (%) of bacteria killed by the effect of antimicrobial drugs. This can be estimated from the number of sequencing reads. VBD index is calculated by (1 − (S × R₀)/(R × S₀)) × 100, where R and S are the numbers of reads from drug-resistant and drug-sensitive bacteria when antibiotics were used, and R₀ and S₀ are the numbers of reads from drug-resistant and drug-sensitive bacteria when no drug was used.