Porechop

FASTQ files were generated for each run using MinKNOW (version 18.2.9; Oxford Nanopore Technologies Ltd.) and demultiplexed with Porechop (32 (link)) and qcat (version 1.01; Oxford Nanopore Technologies Ltd.). Reads were mapped to the M. tuberculosis H37Rv reference strain (GenBank accession number AL123456.3) with minimap2 (33 (link)), and consensus sequences and variants for each sample were generated using samtools (25 (link)) and custom in-house Perl scripts (available on request). Alignments were inspected using AliView (34 (link)), and statistical analysis was performed in R (35 ).

Library preparation and sequencing were performed at the GeT‐PlaGe core facility, INRA Toulouse, according to the manufacturer's instructions “1D gDNA selecting for long reads (SQK‐LSK109).” Aiming at covering the M. graminicola genome at >70× with long reads, sequencing was done on one ONT flowcell. Genomic DNA was purified using AMPure XP beads (Beckman Coulter). Eight micrograms of purified DNA was sheared at 20 kb using the megaruptor system (Diagenode). A “one‐step” DNA damage repair + END‐repair + dA tail of double‐stranded DNA fragments was performed on 2 µg of DNA. Adapters were ligated to the library that was then loaded (0.03 pmol) onto an R9.4.1 revD flowcell. It was sequenced on the GridION instrument for 48 hr. Final reads were base‐called using Guppy v.1.8.5‐1 (Oxford Nanopore).
After sequencing, adapters of raw ONT reads were trimmed using Porechop (Wick, 2019). Only reads with a Q‐score value greater or equal to 7 were selected using NanoFilt v.1.1.0 (De Coster, D’Hert, Schultz, Cruts, & Van Broeckhoven, 2018). Minimap2 (Li, 2018) was used to map long reads to the M. graminicola mitogenome (GenBank no. HG529223), and Samtools Fasta ‐ f 0x4 (Li et al., 2009) was used to sort out long reads that mapped to this reference.