rsomics-methyl

rsomics-methyl is the rsomics product for bisulfite-sequencing methylation extraction and bias QC. It consolidates the team-owned historical rsomics-methyldackel implementation into one product family with MethylDackel compatibility gates.

The command tree exposes checked CpG/CHG/CHH extraction with context-specific transactional bedGraph or methylKit outputs, exhaustive cytosine reports, positional M-bias TSV/SVG reports, standalone context merging, and per-alignment CpG metrics. Standard counts, fractions, total depth, logit, and methylKit representations are available. Complementary CpG and CHG strand calls can be merged before minimum-depth filtering. Indexed 1-based inclusive region selection is shared by extraction, M-bias, and per-read reporting. Extraction and M-bias share strand- and read-specific inclusion bounds and fixed-end trimming. All three operations accept plain or gzip-compressed BED selection. With --keep-bed-strand, BED + selects top-strand and - bottom-strand evidence; per-read selection uses the complete alignment span. Extraction and M-bias also share read-level non-CpG conversion-efficiency filtering. Opposite-strand depth and non-reference fractions can exclude likely variant cytosines during extraction, including their complete merged CpG or CHG context.

rsomics-methyl extract reference.fa alignments.bam --output-prefix sample --region chr1:1-1000000
rsomics-methyl mbias reference.fa alignments.bam --output-prefix sample
rsomics-methyl extract reference.fa alignments.bam --output-prefix selected --bed targets.bed --keep-bed-strand
rsomics-methyl extract reference.fa alignments.bam --output-prefix trimmed --OT 5,100,1,96
rsomics-methyl extract reference.fa alignments.bam --output-prefix sample --format cytosine-report --chg --chh
rsomics-methyl merge-context reference.fa sample_CpG.bedGraph --output merged.bedGraph
rsomics-methyl per-read reference.fa alignments.bam --output reads.tsv

License

Licensed under either Apache License 2.0 or MIT, at your option. MethylDackel behavior provenance and adapted-code attribution are recorded in THIRD_PARTY.md.