v1.1.1

Bug fixes

• Count-matrix flag defaults: --group_novel and --save_mtx now default to on, --save_csv to off. Previously all three silently defaulted to off because the paired argparse flags shared a dest without an explicit default, so a run without explicit flags wrote no count matrix at all, contradicting the documented behavior. Explicit flags (--group_novel_off, --save_mtx_false, --save_csv) still override the defaults.
• --group_novel_off crash: the count matrix task no longer raises a TypeError when novel-isoform grouping is disabled; the annotation pickle is now always loaded for novel-isoform bookkeeping.
• Auxiliary spelling: outputs are written to auxiliary/; legacy auxillary/ folders from older runs are still read.

Documentation

• README updated to match the new count-matrix defaults, and clarified SCOTCH capabilities.
• Fixed an unbalanced parenthesis in the scotch_gene example.

v1.1.0

Performance

• Speed up annotation by skipping GTF parsing when a pkl is present and parallelizing by chromosome
• Replace per-base coverage dictionaries with NumPy difference arrays
• Parallelize BAM info extraction by chromosome and vectorize it
• Cache BAM handles across metagenes and defer GTF loading in the compatibility matrix step
• Add SQLite-backed lookup dictionaries to lower memory in the compatibility matrix step (--save_mem, on by default)

Bug fixes

• Fix gene ID parsing for versioned and non-human identifiers
• Fix empty count matrices for genes with non-Ensembl identifiers
• Fix gene name truncation when splitting spliced and unspliced matrices
• Fix known-isoform classification for non-Ensembl transcript IDs
• Fix per-gene mapping file discovery for non-human builds
• Fix database-locked error when parallel array jobs convert pkl to SQLite
• Fix data loss during pkl-to-SQLite conversion
• Fix per-file SQLite and pkl loading, and add stale-file detection
• Fix edge cases in parallel BAM extraction
• Fix read_selection.pkl rebuild when the file is absent or empty
• Fix MappingScore isoform assignment not reaching the count matrix
• Fix novel annotation routing for single-job runs (#14)

Documentation

• Correct the documented default of --novel_read_n (0, not 20)
• Fix the publication link and citation format in the README

SCOTCH v1.0.2

Minor update to improve speed

SCOTCH v1.0.1

Update the citation.

SCOTCH v1.0.0

SCOTCH v1.0.0