CroCoDeEL v1.3.1

Performance

• Contamination search is about 4.5 times faster: on a 237-sample dataset it now takes 12.5 seconds instead of 58 seconds using 8 CPUs, with identical results

Bug fixes

• easy_wf again exposes only the three mandatory files (-s, -c, -r); use search_conta and plot_conta to change any other setting
• Fixed scatterplots being wrong when a second abundance table is provided: species absent from one table were drawn at 100% abundance instead of being treated as absent
• Made contamination event ordering deterministic, so identical inputs always produce identical output files
• Contamination rates of zero are no longer reported or accepted
• No PDF report is written when no contamination is found, instead of an unreadable empty file
• --nproc now defaults to the CPUs actually available to the process, rather than every core on the machine, and an existing OMP_NUM_THREADS setting is preserved

Improved error handling

• Malformed species abundance tables (empty or with an inconsistent number of columns) now report a clear message instead of a traceback
• Malformed contamination event files are rejected with the offending line number, including rows with missing or extra fields
• Contamination event files that name samples absent from the abundance table are rejected before the report is generated
• Invalid Random Forest model files are rejected before the search starts
• Species abundance tables containing missing values are rejected
• Error messages listing many sample names are truncated instead of printing thousands of names on one line

Other

• Replaced scikit-learn's RANSAC with a specialised implementation in a dedicated module, giving identical results
• Added mypy and pylint to the continuous integration workflow
• Added Python 3.13 to the test matrix and restricted supported versions to 3.12–3.13
• Added unit tests, bringing coverage to 98%
• Refactored the command-line interface, packaged-resource handling and execution description

CroCoDeEL v1.3.0

Bug fixes

• Replaced abrupt sys.exit calls with proper exception handling for input-data errors
• Improved validation of contamination event files to provide clearer error messages for malformed input files
• Improved CLI error handling and exit codes

Other

• Refactored plot_conta, search_conta, and train_model modules
• Added extensive unit tests for the main modules and CLI
• Added input validation and type checking throughout the codebase
• Added documentation and missing docstrings
• Renamed the test_install command to self_test

CroCoDeEL v1.2.2

Update README and improve CroCoDeEL log messages.

CroCoDeEL v1.2.1

Bug fixes

• Switched output format from raw numbers to scientific notation to improve handling of low-rate contamination events.
• Updated rounding logic: replaced fixed decimal rounding with dynamic, magnitude-based rounding.

CroCoDeEL v1.2.0

Bug fixes

• Fixed a rare bug where RANSAC fails to find a valid consensus set
• Fixed incorrect score function in the linear model

New features

• Added a function to refine contamination line detection using Tukey's technique

Other

• Updated the Random Forest model with more relevant features

CroCoDeEL v1.1.1

Improvements

• Refactored and simplified the linear regression implementation
• Vectorized the search for candidate species forming a contamination line
• Limited the maximum number of trials in the RANSAC procedure

Performance

• Achieved ~65% reduction in total runtime (~3× speedup) 🚀

CroCoDeEL v1.1.0

Improvements

• Refactored the codebase for improved structure and maintainability
• Optimized classification performance by reducing the Random Forest size from 1000 to 100 trees

Performance

• Achieved ~35% reduction in total runtime (~1.5× speedup) 🚀

CroCoDeEL v1.0.8

Replace deprecated argparse.FileType to support Python >= 3.14

CroCoDeEL v1.0.7

Add CLI parameters to customize contamination reporting cutoffs
Add a new subcommand (train_model) to train a custom Random Forest classifier
Slight performance improvements
Improve documentation and logging messages

CroCoDeEL v1.0.6

Add a feature to filter low-abundance species and enhance CroCoDeEL's sensitivity with taxonomic profilers other than Meteor.
Fix a bug when merging two abundance tables containing a different number of species. #2
Improve logging messages and update the documentation.