Be notified of new releases
Create your free GitHub account today to subscribe to this repository for new releases and build software alongside 40 million developers.Sign up
This release includes mostly bugfixes and stability improvements. However, one change is potentially braking: subcontigs are now labeled with the suffix
XX is the part number. This is enforced by the core script used to create the coverage table.
For a complete list of changes, see below:
- #236 - Always add suffix to contigs at cutup, even when they are not cut.
- #254 - Slight cleanup of concoct refine
- #258 - New suffices (
.concoct_part_XX) are now used for contig parts
- #261 - Epsilon argument removed as it was not working and is not very useful
- #262 - Rewrote documentation, including installation instructions
- #264 -
concoct_part_suffix is enforced in subcontig for coverage script
- #264 - Header line is enforced for input for
- #267 - Updated documentation
- #253 - A dockerfile useful to test the conda installation
- #258 - Tests for all fundamental scripts, including a new integration test data repository
- #259 - This changelog
- #262 - Added documentation for the core scripts used with concoct
- #265 - A warning is now printed when concoct runs in single threaded mode
- #230 - Enable at least single threaded installation on Mac OSX
- #231 - Replace pandas .ix with .loc to fix deprecation warnings
- #246 - Limit some dependency version numbers for python 2
- #254 - Concoct refine now works with python 3
- #258 - Seed tests now working again
- #260 - Fix the dockerfile build by adding integration test data
- Improved parallelism using OpenMP allowing efficient scaling to a large number of parallel threads
- By only running one instead of ten iterations of the same algorithm, speed is improved drastically.
As an example: A test run on real data resulted in a 22x speed up on a 20 core computer, from 15hours 28min to 42min.
- Now runs on both Python 2 and Python 3. - Please beware that the performance on Python 3 is slightly faster than python 2.
- Installation through conda is now supported:
conda install concoct
- A new step
concoct_refineis introduced where one can rerun the concoct algorithm on seemingly merged clusters.
- Possible to run without rerunning mapping against cut up contigs. This improves the compatibility of concoct against other existing binning software.