# 1.4.1
- count: add `--umi-len` to truncate the UMI to its first N nt, dropping low-quality
  read-through cycles past the designed UMI that would otherwise inflate unique-UMI counts.

# 1.4.0
- demux: add `crisprbact demux-sheet` to generate a bcl2fastq / bcl-convert sample sheet
  and bases mask (handling the umi1-inside-the-index layout) directly from the index CSV.
- demux: add `crisprbact count` to build the guide × UMI count matrix from per-sample FASTQ
  already demultiplexed by the converter (umi reads or umi-in-header). Recovers the full
  umi1+umi2 UMI instead of being limited to whatever the delivered FASTQ contained.
- The original `crisprbact demux` (header-based, single multiplexed file) is kept as a
  fallback for delivered data without BCLs.

# 0.3.11
- Remove fully matched targets from smaller seeds

# 0.3.10
- Add results for fully matched targets

# 0.3.9
- Fix bug #11 (off-target sequence uppper)

# 0.3.8
- Fix bug shift one nucleotide for guide sequence
- Add guide positions on input target sequence

# 0.3.7
- Fix bug displaying wrong guide sequence

# 0.3.6
- Handle genome assembly with contigs

# 0.3.5
- Add longest perfect match between the guide and the off-target sequence to the results

# 0.3.4
- Raise exception when off-target sequence format is wrong

# 0.3.3
- Compute prediction for seeds 8-12
- Prefix off-target feature key result with "off_target"
- Change header of output format
- Change some of the command lines : -g => -s and -gf => -w

# 0.3.2
- Fix bugg when no off-target feature
- Handle genome in fasta file

# 0.3.1
- Fix bug when no feature
- Add off-target strand to output

# 0.3.0
- on_target_predict return an item per feature

# 0.2.1
- Can set seed size

# 0.2.0
- Compute off-target

# 0.1.0
- Initial Release
