- Changed datatype for data simulation back to
float32
- Fixed bugs in scaden model definition [#88]
- removed installation instructions for bioconda as not functional at the moment [#86]
- Fixed bug in
scaden example[#85]
- Reduced memory usage of
scaden simulatesignificantly by performing simulation for one dataset at a time. - Using
.h5adformat to store simulated data - Allow reading data in
.h5adformat for improved performance (courtesy of @eboileau) - Improved logging and using rich progress bar for training
- Gene subsetting is now done only when merging datasets, which will allow to generate different combinations of simulated datasets
- Added
scaden mergecommand which allows merging of previously created datasets
- General improvement of logging using the 'rich' library for colorized output
- Added verification check for '--train_datasets' parameter to notify user of unavailable datasets
- Made identification of datasets more robust to fix issue #66
- Rebuild Scaden model and training to use TF2 Keras API instead of the old compatibility functions
- added
scaden examplecommand which allows to generate example data for test-running scaden and to inpstec the expected file format - added more tests and checks input reading function in
scaden simulate - fixed bug in reading input data
- fixed Dockerfile (switched to pip installation)
- added better error messages to
simulatecommand - cleaned up dependencies
- added
--seedparameter to allow reproducible Scaden runs - added
scaden simulatecommand to perform bulk simulation and training file creation - changed CLI calling
- fixed dependencies (added python>=3.6 requirement)
- upgrade to tf2
- cleaned up dependencies
- small code refactoring
- RAM usage improvement
- added automatic removal of duplicate genes
- changed name of prediction file
Initial release of the Scaden deconvolution package.
Commands:
scaden process: Process a training dataset for trainingscaden train: Train a Scaden modelscaden predict: Predict cell type compositions of a given sample