v0.1.13

Fixes

• [extract, mod_bam] Potential stack overflow when many reads are skipped.
• [extract] Output inferred canonical calls when the . MM tag is used.
• [adjust-mods] change MM flag to ? and add inferred canonical calls when --edge-filter is applied.

Adds

• [pileup] performs pileup in batches to decrease peak memory usage, configurable with --chunk_size.
• [pileup] Allow users to narrow output to any number of specified motifs (instead of only CpG dinucleotides) with --motif option

Changes

• [extract] adds inferred column

v0.1.13-rc1

Fixes

• [extract, mod_bam] Potential stack overflow when many reads are skipped.

v0.1.12

Fixes

• [extract] no longer requires an index to use --num-reads, it will just take the first --num-reads reads if no index is found.

Adds

• New command, modkit repair allows projecting base modifiction calls from one modBAM onto another where the reads have beem trimmed or clipped.
• adjust-mods, update-tags, sample-probs, summary, call-mods and extract can now be used with standard input/output streams.

v0.1.11

Fixes

• [extract] Correctly handles duplex modBAMs when using --include-bed
• Refactors handling of CRAM when sampling records. Fixes #35. Suggests to use BAM.

Adds

• [extract] Output contains ref_mod_strand and modified_primary_base
• [pileup] New option --max-depth allows increasing depth beyond default in hts-lib and sets the default to 8,000 in case it changes in hts-lib.

v0.1.11-rc1

Fixes

• [extract] Correctly handles duplex modBAMs when using --include-bed

v0.1.10

Fixes

• Don't log message whilst aggregating base modification probabilities (exposed with duplex reads as input).

Changes

• [pileup] Fail fast when bam index doesn't contain any mapped reads.

v0.1.9

Changes

• [pileup] When estimating the pass-threshold only use base modification probabilities if the read base is aligned to the reference (don't use soft-clipped and inserts). Use --include-unmapped to use all base modification probabilities.
• [adjust-mods] requires --ignore, --convert, or --edge-filter.

Adds

• [pileup, extract, sample-probs, summary] Allow narrowing of analysis to specific sites with --include-bed.
• [summary, sample-probs] Add --only-mapped flag that will only report on base modification probabilities if they are mapped to the reference.
• [pileup] Allow partitioning counts to separate bedMethyl files based on SAM tags with --partition-tag option.

Fixes

• [adjust-mods, update-tags, call-mods] Panic when failure to parse SAM header, fixes #29.

v0.1.8

Changes

• [call-mods] Emit 0 ML value when a mod code is not called, previously the call was omitted (only the called mod was emitted).

Adds

• [adjust-mods, call-mods, pileup, sample-probs, summary, extract] Allow --edge-filter that will remove base modification probabilities at the ends of reads.

Fixes

• [call-mods] --no-filtering flag properly handled.

Deprecated

• [adjust-mods] In order to allow --edge-filter without also performing an ignore or convert, in the next release --ignore will require a explicit argument (h is no longer the default). In this release, when no edge filter, conversion, or ignore is provided, h will be used for ignore, but this behavior will be removed in the next release.

v0.1.7

Changes

Adds

• modkit extract sub-command that produces a table of per-read base modification probabilities.
• Allow suppression of progress bar with --suppress-progress

Fixes

• Don't count skipped reads towards total when sampling.

N.B.

modkit call-mods has a bug in this release where --no-filtering is not handled properly (filtering is still performed). Use a later release or as a work-around use --filter-threshold 0.0.

v0.1.6

Changes

• When ignoring a base modification in pileup, calculate pass thresholds on the probabilities after ignoring the specified probability.

Adds

• [pileup] API to specify pass thresholds per base modification (or canonical) with --mod-threshold option.
• New modBAM->modBAM subcommand, call-mods that will apply pass thresholds and clamp base modification probabilities to zero (canonical) or 1.0 for base modifications
• [sample-probs, summary] Allow --ignore to remove a base modification class from the analysis.