Nanopore-seq analysis

The nanopore data was processed with ONT_hg_pipe v.0.1.0 (Palin 2018, unpublished, https://github.com/kpalin/). The GP5d nanopore data was basecalled with Guppy v.5.0.17 with the super-accurate basecalling model and a minimum q-score of 10. The reference genome was indexed and the basecalled reads were aligned to the reference genome with minimap2 v.2.16 (minimap2 –x map-ont)⁸⁸. Quality controls were performed with nanoplot v.1.20.0⁸⁹ and Samtools v.1.9⁷¹. After alignment, methylation was called with nanopolish v.0.11.1⁹⁰. The cpggpc_new_train branch in GitHub (https://github.com/jts/nanopolish/tree/cpggpc_new_train) was used to call both CpG and GpC methylation (nanopolish call-methylation -q cpggpc). The resulting table was processed to a BED format (mtsv2bedGraph.py -q cpggpc—nome) and to methylation frequency table formats for CpG and GpC methylation (parseMethylbed.py frequency -v -m cpg and parseMethylbed.py frequency -v -m gpc), using previously published scripts⁷². The resulting methylation tables were converted to bedGraph and bigwig formats with a custom script (mfreq_to_bw.R), utilizing bedGraphToBigWig v.377⁸⁶. The CpG methylation frequency tables were loaded into R and smoothed with bsseq v.1.28.0 (BSmooth ns = 50, h = 1000, maxGap = 100000)⁹¹. The GpC methylation calling was performed for another project and was not used in this study.