Fix URL to downloadable resources

Fixed a bug that was inhibiting automatic download of CellO's resources files.

Compatibility with sparse matrices

This updates enables CellO to work with sparse matrices.

Minor bug fixes

Minor bug fix and new feature of optionally returning graph object from scanpy_cello.cello_probs function.

Minor bug fix

Fixed a bug in which the cello_predict command is run with the -f option specifying the location of the CellO resources location, but CellO does not recognize that the location of the resources was provided.

Minor bug fix

• Minor bug fix when genes are incompatible between input dataset and training set
• Updates to Jupyter notebook tutorial
• Output file containing all genes that are common between input dataset and training set

Small enhancements.

The Cell Ontology is now loaded lazily into memory rather than during import of the CellO package. This speeds up the import.

Easier installation. Integrates with Scanpy.

This release presents a number of significant changes:

1. A setup.py script enables easy installation
2. CellO can be run using command line tools
3. CellO can be run within Python in conjunction with Scanpy and thus can be integrated into more general single-cell analysis workflows

Fixed issue reading thresholds

Fixed several issues:

• Incorrect thresholds were utilized for calling yes-no cell type decisions
• Problem with the tool reading units other than counts
• Aggregates clusters using the raw counts, as was done in the evaluation in the paper, rather than the CPM's

Fixed bug mapping cells to clusters

Fixed a significant bug in which the cells were mapped to the wrong clusters when generating the output files and thus, each cell's output label was incorrectly assigned. CellO also now outputs a third file that describes the most-specific predicted cell type for each cell.

First release of CellO

First release of CellO