All notable changes and fixes to ont_fast5_api will be documented here
The format is based on Keep a Changelog This project (aspires to) adhere to Semantic Versioning.
- Support for python up to 3.12
- Support for h5py>=3.9
- Renamed VBZ compression settings to make it clearer which version is used in production
- Support for python3.6
- Compatibility with numpy==1.24 unicode type
- Updated Windows VBZ Plugin dll
- Support for fast5_api on macOS-M1
- Fixed Fast5Read import error
- Fixed unresolved reference in
compress_fast5.py
- Fixed issue with
compress_fast5.py
not retaining enumeration metadata for the end_reason attribute - Increased minimum h5py version to 2.10
- Script
demux_fast5
for demultiplexing fast5 reads based on column in summary file, e.g. for barcoded experiments
- Removed deb builds which are no longer supported
- Python3.5 support
- Added
yield_fast5_reads
to conversion_tools.
- Dropped support for older h5py/numpy versions, min now h5py>=2.8, numpy>=1.16
- fast5_subset now displays errors (but continues processing) when it encounters input fast5 files it can't read.
- Add support for explicitly specifying file drivers when loading multi-read fast5 files.
compress_fast5
now has a--sanitize
option to remove optional groups.
- Correctly handle the case where h5pl can be imported but doesn't have the prepend() function available.
- Added explicit requirements and checks to prevent running on Python 2.
- Compression now works in
single_to_multi
.
- Compression argument in
fast5_subset
andsingle_to_multi
failed if not set
- Compression argument in
fast5_subset
andsingle_to_multi
was parsed incorrectly
- Argument list for
fast5_subset
andsingle_to_multi
had a syntax error
- Hardlinking of metadata to prevent duplication and reduce filesize
- Ability to enable compression when using
fast5_subset
andsingle_to_multi
fast5_subset
thread pool could sometimes close before all tasks were completedfast5_subset
will create output directory if it doesn't exist
- Comparison of file_versions could throw an error
- Basecall1DTools could not load data from a Fast5Read
- python2 compatibility
- minor documentation errors: #28
- Script to check the compression type of fast5 files in a folder
compress_fast5
can now be used--in_place
- Reading arrays with padded strings now succeeds (on h5py>2.7)
- Compatibility bugs with h5py==2.6 now raises appropriate errors
- Fast5File now has attribute read_id to match documentation
- Now use standard settings for gzip compression (gzip=1, shuffle=None)
- Inverted dependency between
Fast5File
andFast5Read
soFast5Read
is now the primary object
- Option to
--ignore_symlinks
in fast5 conversion scripts - Explicit check to file_type for detemining single/multi-read files
fast5_subset
with single read fast5s was failing- unit test data now cleaned up properly
- Compatibility for VBZ compressed reads
compress_fast5
script for compressing/decompressing fast5 filesget_reads()
helper method to more easily loop through reads in a fast5 file
Fast5File().get_raw_data()
updated interface to matchFast5Read
and remove support for legacy files with multiple read numbers in a singleFast5File
- Minimum depedency version requirements bumped. Set to Ubuntu16
apt
python3-package defaults
- Legacy
Fast5Writer
object.MultiReadFast5
orEmptyFast5File
are preferred
- Check for progressbar2 package and fail early if it's installed.
- Support for h5py==2.10 string data type encoding changes
- Corrected some "for for" typos in argparse help text.
- Bug in read string and read_id concatenation resulted in broken output file
- Updated fast5_subset script to extract also from single-read fast5 files
- Renamed fast5_subset source script from multi_fast5_subset.py to fast5_subset.py
- Bug in number of processes being 0 when batch size is greater than number of reads (py2)
- Bug in path name output from pathlib changes
- Bug with apt-install and pathlib2
- get_raw_data() now works with scale=True when start,end are None
- Useful error message if no input files found
- filename_mapping output gave incorrect filenames
- Script for extracting reads by id from
multi_read
files
- Bug in output to
filename_mapping.txt
- Multi-threading support for multi<->single conversion for improved performance
- Removed incorrect license accidentally added to README
- Minor documentation updates
- Follow symlinks when finding files recursively
- Generic single- and multi- read interface via
get_fast5_file
- Incorrect time estimates for single-multi conversion
- Fixed path creation if not exist
- Support for multi-read fast5 files
- Conversion tools for single-multi read files
- Support for deprecated interface to Basecall2D following 0.4.0, support will end in v1.x.x
- Basecall1d and Basecall2d raise consistent KeyError when fastq data missing
- Interface to Basecall1d and Basecall2d unified for add_sequence() and get_sequence()
- Fast5 file now supports logging via 'Fast5File.add_log()'
- Invalid component names no longer checked against LEGACY_COMPENENTS
- Raise KeyError when fastq data missing from Basecall1d
- median_before and start_mux populate correctly with sensible defaults
Major release - changes not logged before this point