
# 0.9.0
- Staged everything for pushing to PyPi, including CI
- Merged pull request features, including removing the rmbackground (now in mitobender) and handling small features like gzip'd barcode lists

# 0.8.0
- Removed the `java`/Picard MarkDuplicates dependency entirely; duplicate
  removal now runs through `bam-dedup` (pure Python/Cython, pysam-based).
  `mgatk/bin/picard.jar` (~15MB) is gone from the repo.
- **Breaking**: `--max-javamem`/`-jm` is deprecated and does nothing (there's
  no JVM left to size). It's still accepted so existing scripts don't fail;
  passing it now prints a one-line notice instead.
- Refactored the standalone scripts under `mgatk/bin/python/` (invoked via
  `os.system('python script.py ...')`) into a proper importable package,
  `mgatk.processing`, called directly from the CLI and Snakemake rules.
  Hardened the remaining external-process calls (snakemake, Rscript,
  cellbender) to check for failure instead of silently ignoring it.
- Modernized dependencies for current Python/snakemake/pysam/numpy/pandas;
  fixed several Python-2-only and deprecated-API bugs along the way.
- Replaced the placeholder test suite (a single test with no assertions)
  with real unit and end-to-end coverage; refreshed `tests/README.md` and
  the GitHub issue templates.

# 0.7.1
- Fix various versioning bugs

# 0.7.0
- Fix deprecation error in yaml
- Fix deprecation error in snakemake

# 0.6.9
- UMI parsing fix for `tenx` mode thanks to @yaniv-el

# 0.6.8
- Speedups for `tenx` mode thanks to @ruochiz

# 0.6.7
- Updates to deletion calling module

# 0.6.6
- Fix naming parameters in tags
- update `check` issue with tenx mode

# 0.6.4
- Allow for barcodes with ACGT but not the -n convention

# 0.6.3
- Facilitate CellRanger-ATAC v2 bam file
- Alpha version of remove-background 

# 0.6.2
- Fixed bug associated with tenx mode

# 0.6.1
- Automated variant calling in python thanks to @vincent6liu
- Command line option to handle overlapping mate reads thanks to @vincent6liu

# 0.6.0
- Functionality for mgatk-del

# 0.5.9
- Fix bug in strand assignment in `(b)call` mode

# 0.5.8
- Fix (hopefully innocuous) bug thanks to @bobermayer 

# 0.5.7 
- Automatically generate .rds file for use with Signac

# 0.5.6
- Added `tenx` mode for more efficient processing of large single-cell datasets

# 0.5.0
- Beta version distributed via PyPi
- Revamped code used in mtscATAC-seq paper

# 0.4.0
- Fully separated by strand for counts / qualities
- UMI-based duplicate read removal available
- Default to remove PCR duplicates

# 0.3.9
- Handle common reference genome issue for human genomes (hg19 / rCRS) automatically

# 0.3.4
- Full support for `bcall` mode

# 0.3.0
- Per-base / allele quality scores

# 0.2.5
- Creates .rds for easy i/o into R package
- Efficient sparse matrices for final text-based output
- Added additional run modes. 5 total as of now (call, one, gather, support, check)

# 0.1.0
- Alpha version uploaded to PyPi

# 0.0.1
- Added NEWS

